The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

Review Articles

Medical Progress Pressure-support ventilation also provides graded

assistance but differs from the other two modes in
that the physician sets the level of pressure (rather
A DVANCES IN M ECHANICAL than the volume) to augment every spontaneous res-
V ENTILATION piratory effort.5 The level of pressure delivered by the
ventilator is usually adjusted in accordance with chang-
MARTIN J. TOBIN, M.D. es in the patients respiratory frequency. However, the
frequency that signals a satisfactory level of respiratory-

T
HE chief reason that patients are admitted to muscle rest has never been well defined, and recom-
an intensive care unit is to receive ventilatory mendations range from 16 to 30 breaths per minute.6
support. In this article, I update the basic prin- New modes of mechanical ventilation are often in-
ciples of mechanical ventilation, which I reviewed in troduced. Each has an acronym, and the jargon is in-
the Journal in 1994,1 and discuss recent advances. hibiting to those unfamiliar with it. Yet each new
mode involves nothing more than a modification of
BASIC PRINCIPLES the manner in which positive pressure is delivered to
The indications for mechanical ventilation, as de- the airway and of the interplay between mechanical
rived from a study of 1638 patients in eight countries,2 assistance and the patients respiratory effort. The pur-
are acute respiratory failure (66 percent of patients), pose of a new mode of ventilation may be to enhance
coma (15 percent), acute exacerbation of chronic ob- respiratory-muscle rest, prevent deconditioning, im-
structive pulmonary disease (13 percent), and neu- prove gas exchange, prevent lung damage, enhance
romuscular disorders (5 percent). The disorders in the the coordination between ventilatory assistance and
first group include the acute respiratory distress syn- the patients respiratory efforts, and foster lung heal-
drome, heart failure, pneumonia, sepsis, complications ing; the priority given to each goal varies.
of surgery, and trauma (with each subgroup account-
ing for about 8 to 11 percent of the overall group). COORDINATING RESPIRATORY EFFORT
The objectives of mechanical ventilation are primar- AND MECHANICAL VENTILATION
ily to decrease the work of breathing and reverse life- Probably the most common reason for instituting
threatening hypoxemia or acute progressive respira- mechanical ventilation is to decrease the work of the
tory acidosis. respiratory muscles. The inspiratory effort expended
Virtually all patients who receive ventilatory sup- by patients with acute respiratory failure is about four
port undergo assist-control ventilation, intermittent times the normal value, and it can be increased to six
mandatory ventilation, or pressure-support ventilation; times the normal value in individual patients.7 Criti-
the latter two modes are often used simultaneously.2 cally ill patients in whom this increased level of effort
With assist-control ventilation, the most widely used is sustained indefinitely are at risk of inspiratory-mus-
mode, the ventilator delivers a set tidal volume when cle fatigue, which can add structural injury to already
triggered by the patients inspiratory effort or inde- overworked muscles.8 It is sometimes thought that
pendently, if such an effort does not occur within a the simple act of connecting a patient to a ventilator
preselected time. will decrease respiratory effort. Yet unless the settings
Intermittent mandatory ventilation was introduced are carefully selected, mechanical ventilation can ac-
to provide graded levels of assistance. With this mode, tually do the opposite.
the physician sets the number of mandatory breaths With careful selection of ventilator settings, in-
of fixed volume to be delivered by the ventilator; be- spiratory effort can be reduced to the normal range.9
tween these breaths, the patient can breathe sponta- But eliminating inspiratory effort is not desirable be-
neously.3 Patients often have difficulty adapting to cause it causes deconditioning and atrophy of the res-
the intermittent nature of ventilatory assistance, and piratory muscles.10 Surprisingly, researchers have not
the decrease in the work of breathing may be much attempted to determine the desirable target for reduc-
less than desired.4 ing inspiratory effort in patients with acute respira-
tory distress. To reduce effort markedly requires that
From the Division of Pulmonary and Critical Care Medicine, Edward
the ventilator cycle in unison with the patients central
Hines, Jr., Veterans Affairs Hospital and Loyola University of Chicago respiratory rhythm (Fig. 1). For perfect synchroniza-
Stritch School of Medicine, Hines, Ill. Address reprint requests to Dr. To- tion, the period of mechanical inflation must match
bin at the Division of Pulmonary and Critical Care Medicine, Edward
Hines, Jr., Veterans Affairs Hospital, Rte. 111N, Hines, IL 60141, or at the period of neural inspiratory time (the duration
[email protected]. of inspiratory effort), and the period of mechanical

1986 N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org
 MED ICA L PROGR ES S

120

liters per minute

0 Flow

120
0 1 2 3 4 5 6 7
20
cm of water

10
Airway
0 pressure
10
0 1 2 3 4 5 6 7

Diaphragm

Transversus
abdominis

0 1 2 3 4 5 6 7
Seconds

Figure 1. Flow, Airway Pressure, and Inspiratory and Expiratory Muscle Activity in a Patient with Chronic Obstructive Pulmonary
Disease Who Received Pressure-Support Ventilation at an Airway Pressure of 20 cm of Water.
The electromyograms in the lower portion of the figure show inspiratory muscle activity in the patients diaphragm and expiratory
muscle activity in the transversus abdominis. The patients increased inspiratory effort caused the airway pressure to fall below the
set sensitivity (2 cm of water), and inadequate delivery of flow by the ventilator resulted in a scooped contour on the airway-pres-
sure curve during inspiration. While the ventilator was still pumping gas into the patient, his expiratory muscles were recruited,
causing a bump in the airway-pressure curve. That the flow never returned to zero throughout expiration reflected the presence of
autopositive end-expiratory pressure. The broken red line shows airway pressure in another patient, who generated just enough
effort to trigger the ventilator and in whom there was adequate delivery of gas by the ventilator. Data are from Jubran et al.6 and
Parthasarathy et al.11

inactivity must match the neural expiratory time.12,13 often, the clinician sets a threshold for airway pres-
Difficulties in synchronization can arise at the onset sure that will trigger the ventilator. This threshold,
of inspiratory effort, at the onset of flow delivered by referred to as set sensitivity, is usually 1 to 2 cm
the ventilator, during the period of ventilator-induced of water.14 To reach this threshold, the patient must
inflation, and at the switch between inspiration and initiate an inspiratory effort. But when the threshold
expiration. is reached, inspiratory neurons do not simply switch
Almost all patients who undergo mechanical ven- off. Consequently, the patient may expend consider-
tilation receive some form of assisted ventilation, with able inspiratory effort throughout the machine-cycled
the patients inspiratory effort triggering the ventila- inflation.15
tor. To ensure that the ventilator does not cycle too The display of airway pressure and flow tracings

N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org 1987
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

on ventilator screens has increased awareness that in- ly through an increase in airway pressure. Satisfactory
spiratory effort is frequently insufficient to trigger oxygenation is easily achieved in most patients with
the ventilator. At high levels of mechanical assistance, airway obstruction. The main challenge arises in pa-
up to one third of a patients inspiratory efforts may tients with alveolar-filling disorders, especially the
fail to trigger the machine.9,16,17 Surprisingly, unsuc- acute respiratory distress syndrome a form of non-
cessful triggering is not the result of poor inspiratory cardiogenic pulmonary edema resulting from severe
effort; indeed, the effort is more than a third greater acute alveolar injury. It has long been recognized that
when the threshold for triggering the ventilator is arterial oxygenation can be achieved at a lower in-
not reached than when it is reached.9 Breaths that do spired oxygen concentration by increasing airway pres-
not reach the threshold for triggering the ventilator sure. The goal of using the lowest possible oxygen
have higher tidal volumes and shorter expiratory times concentration to achieve an arterial oxygen satura-
than do breaths that do trigger the ventilator. Con- tion of approximately 90 percent has not changed in
sequently, elastic-recoil pressure builds up within the decades. What has changed is how this goal is viewed
thorax in the form of intrinsic positive end-expira- in relation to other factors, particularly ventilator pres-
tory pressure (PEEP), or auto-PEEP.9 To trigger the sures. In recent years, there has been a growing tend-
ventilator, the patients inspiratory effort first has to ency to be more concerned about high airway pres-
generate a negative intrathoracic pressure in order to sures than about oxygen toxicity, although this shift
counterbalance the elastic recoil and then must reach has been based on a consensus of opinion rather than
the set sensitivity. The consequences of wasted in- on data from studies in patients and animals.
spiratory efforts are not fully known, but they add From the outset, clinicians recognized that mechan-
an unnecessary burden in patients whose inspiratory ical ventilation could rupture alveoli and cause air
muscles are already under stress. leaks.20 In 1974, Webb and Tierney showed that me-
The inspiratory flow rate is initially set at a default chanical ventilation could also cause ultrastructural
value, such as 60 liters per minute. If the delivered injury, independently of air leaks.21 Their observa-
flow does not meet the patients ventilatory needs, tions went largely unnoticed until a decade later, when
inspiratory effort will increase.15 Sometimes the flow several investigators confirmed and extended them. Al-
is increased in order to shorten the inspiratory time veolar overdistention causes changes in epithelial and
and increase the expiratory time, especially in patients endothelial permeability, alveolar hemorrhage, and
with inspiratory efforts that are insufficient to trig- hyaline-membrane formation in laboratory animals.22
ger the ventilator. But an increase in flow causes im- Diffuse infiltrates on chest radiographs originally
mediate and persistent tachypnea, and as a result, the led clinicians to infer that lung involvement was ho-
expiratory time may be shortened.18 In one study, mogeneous. But computed tomography (CT) reveals
for example, increases in inspiratory flow from 30 liters a patchy pattern: about one third of the lung is unaer-
per minute to 60 and 90 liters per minute caused in- ated, one third poorly aerated, and one third normal-
creases in the respiratory rate of 20 and 41 percent, ly aerated.23,24 A ventilator-induced breath will follow
respectively.19 the path of least impediment, travelling preferential-
In studies of interactions between the patients ly to the normally aerated areas. As a result, these re-
respiratory effort and mechanical ventilation, remark- gions are vulnerable to alveolar overdistention and the
ably little attention has been paid to the switch be- type of ventilator-induced lung injury found in lab-
tween inspiration and expiration. With the use of pres- oratory animals 25 (Fig. 2).
sure-support ventilation, ventilatory assistance ceases A new era of ventilatory management began in
when the patients inspiratory flow falls by a preset 1990, when Hickling et al.26 reported that lowering
amount (e.g., to 25 percent of the peak flow).5 Air the tidal volume caused a 60 percent decrease in the
flow changes more slowly in patients with chronic expected mortality rate among patients with the acute
obstructive pulmonary disease than in other patients, respiratory distress syndrome. In a subsequent trial,
and patients often start to exhale while the ventilator Amato et al.27,28 randomly assigned patients to a con-
is still pumping gas into their chests.6,11 In 5 of 12 ventional tidal volume (12 ml per kilogram of body
patients with chronic obstructive pulmonary disease weight) or to a low tidal volume (less than 6 ml per
who were receiving pressure support of 20 cm of kilogram). Mortality was decreased by 46 percent
water, expiratory muscles were recruited during ven- with the lower tidal volume. In a recent study of 861
tilator-induced inflation.6 patients, the Acute Respiratory Distress Syndrome
Network 29 confirmed this benefit: mortality was de-
IMPROVING OXYGENATION AND creased by 22 percent with a tidal volume of 6 ml per
PREVENTING LUNG INJURY kilogram as compared with a tidal volume of 12 ml
A primary goal of mechanical ventilation is to im- per kilogram. Lowering the tidal volume, however,
prove arterial oxygenation. Improvement is achieved failed to improve the outcome in three controlled
partly through the use of endotracheal intubation to trials.30-32 The discrepant findings can be explained by
ensure the delivery of oxygen to the airway and part- differences in trial design. Increased survival was de-

1988 N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org
 MED ICA L PROGR ES S

A B

R L R L

C D

Right Lung Left Lung

Figure 2. Lung Injury Caused by Mechanical Ventilation in a 31-Year-Old Woman with the Acute Respiratory Distress Syndrome Due
to Amniotic-Fluid Embolism.
The patient had undergone mechanical ventilation for eight weeks with tidal volumes of 12 to 15 ml per kilogram of body weight,
peak airway pressures of 50 to 70 cm of water, positive end-expiratory pressures of 10 to 15 cm of water, and a fractional inspired
oxygen concentration of 0.80 to 1.00 in order to achieve a partial pressure of carbon dioxide that was less than 50 mm Hg and a
partial pressure of oxygen that was 80 mm Hg or higher. Computed tomography (CT) performed two days before the patient died
revealed a paramediastinal pneumatocele in the right lung (Panel A, arrowheads) and numerous intraparenchymal pseudocysts in
the left lung (Panel B, black arrow, open circle, and asterisk).
At autopsy, both lungs were removed and fixed by intrabronchial infusion of formalin, alcohol, and polyethylene glycol at an insuf-
flation pressure of 30 cm of water. Panel C shows the paramediastinal pneumatocele in the right lung (arrowheads); the horizontal
broken line is the level of the CT section. Panel D shows a 1-cm-thick section of the left lung, corresponding to the CT section. Small
pseudocysts are present (arrow), and two large pseudocysts (asterisk and open circle) have compressed and partially destroyed
the parenchyma. The development of these lesions in a patient without a history of chronic lung disease indicates the harm that
may result with the use of high tidal volumes and airway pressures. The photographs were kindly provided by Dr. Jean-Jacques
Rouby, Hpital de la PitiSalptrire, Paris.

N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org 1989
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

monstrable only when the patients undergoing con- mocapnia represents a substantial shift in ventilatory
ventional ventilation had a mean pressure during an management.
end-inspiratory pause (the so-called plateau pressure, Lowering the tidal volume is not without hazards.
a surrogate for peak alveolar pressure) that exceeded In addition to the potential harm of hypercapnia,
32 cm of water.33 the volume of aerated lung may be decreased,39 with a
The pressures pertinent to ventilatory management consequent increase in shunting and worsening oxy-
are the peak inspiratory pressure, plateau pressure, genation. One means of minimizing the loss of lung
and end-expiratory pressure. Patients with airway ob- volume is the use of sighs (i.e., single breaths of large
struction may have a very high peak pressure without tidal volume). In one study, increasing the plateau
any increase in the plateau pressure. Indeed, the gra- pressure by at least 10 cm of water during sighs, ap-
dient between the two is directly related to the resist- plied three times a minute over a period of one hour,
ance of the airway to airflow. An increase in the peak caused a 26 percent decrease in shunting, with a 50
inspiratory pressure without a concomitant increase percent increase in the partial pressure of oxygen.40
in the plateau pressure is unlikely to cause alveolar It is unknown whether sighs used at this low frequen-
damage. The critical variable is not airway pressure it- cy cause injury from alveolar overdistention.
self but transpulmonary pressure airway pressure The more usual way of improving oxygenation is
during the end-inspiratory pause minus pleural pres- through the use of PEEP with the intention of re-
sure. The normal lung is maximally distended at a cruiting previously nonfunctioning lung tissue. Se-
transpulmonary pressure between 30 and 35 cm of lecting the right level of PEEP for a given patient with
water, and higher pressures cause overdistention. Pa- the acute respiratory distress syndrome is difficult,
tients with stiff chest walls, such as those with the because the severity of injury varies throughout the
acute respiratory distress syndrome due to a nonpul- lungs. PEEP can recruit atelectatic areas but may over-
monary disorder (e.g., abdominal sepsis), have an el- distend normally aerated areas.41,42 In a study involv-
evated pleural pressure. In such patients, the airway ing six patients with acute lung injury, for example, the
plateau pressure may exceed 35 cm of water without use of PEEP at 13 cm of water resulted in the re-
causing alveolar overdistention. cruitment of nonaerated portions of lung, with a gain
Clinical decisions based on plateau pressure must of 320 ml in volume, but three patients had over-
take into account the relation between lung volume distention of already aerated portions of lung, with
and airway pressure in the individual patient. The pres- an excess volume of 238 ml.43
surevolume curve in patients with the acute respi- Overall, about 30 percent of patients with acute
ratory distress syndrome typically has a sigmoid shape lung injury do not benefit from PEEP or have a fall in
with two discrete bends, called inflection points (Fig. the partial pressure of oxygen.23,44,45 With the patient
3).34 Some investigators believe that a plateau pres- in the supine posture, PEEP generally recruits the re-
sure above the upper bend causes alveolar overdisten- gions of the lung closest to the apex and sternum.23
tion. Reducing the tidal volume lowers the plateau Conversely, PEEP can increase the amount of nonaer-
pressure, but at the cost of hypercapnia. In a study in ated tissue in the regions close to the spine and the
which 25 patients with the acute respiratory distress diaphragm.23 Among patients in the early stages of
syndrome underwent mechanical ventilation with a the acute respiratory distress syndrome, those with
tidal volume of 10 ml per kilogram, 20 had a plateau pulmonary causes, such as pneumonia, are less likely
pressure that was 2 to 14 cm of water above the up- to benefit from PEEP than are those with nonpul-
per bend of the pressurevolume curve.35 Lowering monary causes, such as intraabdominal sepsis or ex-
the plateau pressure to a value that fell below the up- trathoracic trauma.46 This distinction may be related
per bend required a 22 percent decrease in the tidal to the type of morphologic involvement: pulmonary
volume, causing the partial pressure of carbon diox- causes of the syndrome are characterized by alveolar
ide to increase from 44 to 77 mm Hg.35 The partial filling, whereas nonpulmonary causes are characterized
pressure of carbon dioxide, in turn, can be decreased by interstitial edema and alveolar collapse. In the lat-
by as much as 28 percent by removing tubing and er stages of the acute respiratory distress syndrome,
thus decreasing dead space and increasing the fre- remodeling and fibrosis may eliminate this distinction
quency of ventilator-induced breaths. By virtue of between pulmonary and nonpulmonary causes.
their stiff lungs, patients with the acute respiratory To select the right level of PEEP, some experts
distress syndrome who do not have an underlying air- recommend bedside calculation of the pressurevol-
way obstruction can tolerate a frequency of 30 breaths ume curve. With the ventilators currently used in the
per minute without gas trapping.36 Severe hypercap- United States, calculating the pressurevolume curve
nia can have adverse effects, including increased in- is logistically difficult and technically demanding.34
tracranial pressure, depressed myocardial contractility, Yet many ventilators have a computer screen, and mi-
pulmonary hypertension, and depressed renal blood nor software modifications would make it feasible to
flow.37,38 The view that these risks are preferable to calculate the curve in as little as two minutes as
the higher plateau pressure required to achieve nor- with the ventilators available in France.47 Providing

1990 N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org
 MED ICA L PROGR ES S

Conventional Ventilation
Alveolar
overdistention

Protective Ventilation
1000

800 800

600 Tidal Volume (ml) 600 600

400 400 400

200 200 200

0 0 0
0 2 4 6 0 10 20 30 40 0 2 4 6
Seconds Pressure (cm of water) Seconds

Alveolar
collapse

Figure 3. Respiratory PressureVolume Curve and the Effects of Traditional as Compared with Protective Ventilation in a 70-kg Pa-
tient with the Acute Respiratory Distress Syndrome.
The lower and upper inflection points of the inspiratory pressurevolume curve (center panel) are at 14 and 26 cm of water, respec-
tively. With conventional ventilation at a tidal volume of 12 ml per kilogram of body weight and zero end-expiratory pressure (left-hand
panel), alveoli collapse at the end of expiration. The generation of shear forces during the subsequent mechanical inflation may tear
the alveolar lining, and attaining an end-inspiratory volume higher than the upper inflection point causes alveolar overdistention. With
protective ventilation at a tidal volume of 6 ml per kilogram (right-hand panel), the end-inspiratory volume remains below the upper
inflection point; the addition of positive end-expiratory pressure at 2 cm of water above the lower inflection point may prevent alveolar
collapse at the end of expiration and provide protection against the development of shear forces during mechanical inflation.

this option on ventilators would increase clinicians pressurevolume curve and events at the alveolar level
experience with the use of pressurevolume curves is confounded by numerous factors and is the sub-
in ventilatory management. ject of ongoing research and debate.51-55 An under-
Even if the pressurevolume curve is not calculated standing of this relation is also impeded by the dif-
at the bedside, it is useful to select the PEEP level ac- ficulty in distinguishing collapsed lung units from
cording to this conceptual framework. A level above fluid-filled units on CT.
the lower bend in the pressurevolume curve is Most patients with the acute respiratory distress
thought to keep alveoli open at the end of expiration syndrome have an increase in the partial pressure of
and thus prevent the injury that can result from shear oxygen when there is a change from the supine to
forces created by the opening and closing of alveo- the prone position. In a study of 16 patients, for ex-
li.48-50 This level of PEEP may also prevent an in- ample, 12 had an increase of 9 to 73 mm Hg in the
crease in the amount of nonaerated tissue and, thus, partial pressure of oxygen, and 4 had a decrease of
atelectasis. However, the notion that the lower bend 7 to 16 mm Hg.56 The mechanism responsible for
signals the level of PEEP necessary to prevent end- the improvement in the partial pressure of oxygen is
expiratory collapse and that pressures above the upper not clear. The attribution of this improvement to lung
bend signal alveolar overdistention is a gross over- recruitment has not been proved.56 It is now posited
simplification. The relation between the shape of the that a prone position causes ventilation to be distrib-

N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org 1991
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

uted more evenly to the various regions of the crease in oxygen extraction, and these patients have a
lungs,57,58 improving the relation between ventilation relative decrease in oxygen delivery.64 The greater ox-
and perfusion.59,60 ygen extraction causes a substantial decrease in mixed
venous oxygen saturation, contributing to the arte-
DISCONTINUING MECHANICAL rial hypoxemia that occurs in some patients.64
VENTILATION Over the course of a trial of spontaneous breath-
Because mechanical ventilation can have life-threat- ing, about half of patients in whom the trial fails have
ening complications, it should be discontinued at the an increase in carbon dioxide tension of 10 mm Hg
earliest possible time. The process of discontinuing or more.7 The hypercapnia is not usually a conse-
mechanical ventilation, termed weaning, is one of the quence of a decrease in minute ventilation.63 Instead,
most challenging problems in intensive care, and it hypercapnia results from rapid, shallow breathing,
accounts for a considerable proportion of the work- which causes an increase in dead-space ventilation. In
load of staff in an intensive care unit.2 a small proportion of patients who cannot be weaned,
When mechanical ventilation is discontinued, up primary depression of respiratory drive may be re-
to 25 percent of patients have respiratory distress se- sponsible for the hypercapnia.7
vere enough to necessitate the reinstitution of venti- The discontinuation of mechanical ventilation needs
latory support.61,62 Our understanding of why weaning to be carefully timed. Premature discontinuation plac-
fails in some patients has advanced considerably in re- es severe stress on the respiratory and cardiovascular
cent years. Among patients who cannot be weaned, systems, which can impede the patients recovery.
disconnection from the ventilator is followed almost Unnecessary delays in discontinuation can lead to a
immediately by an increase in respiratory frequency host of complications. Decisions about timing that
and a fall in tidal volume that is, rapid, shallow are based solely on expert clinical judgment are fre-
breathing63 (Fig. 4). As a trial of spontaneous breath- quently erroneous.68-70 Several functional measures are
ing is continued over the next 30 to 60 minutes, the used to aid decision making. The level of oxygenation
respiratory effort increases considerably, reaching more must be satisfactory before one attempts to discon-
than four times the normal value at the end of this tinue mechanical ventilation. Yet in many patients with
period.7 The increased effort is mainly due to wor- satisfactory oxygenation, such attempts fail. The use
sening respiratory mechanics. Respiratory resistance of traditional predictors of the success or failure of
increases progressively over the course of a trial of attempts maximal inspiratory pressure, vital ca-
spontaneous breathing, reaching about seven times pacity, and minute ventilation frequently has false
the normal value at the end of the trial; lung stiffness positive or false negative results.71 A more reliable
also increases, reaching five times the normal value; predictor is the ratio of respiratory frequency to tidal
and gas trapping, measured as auto-PEEP, more than volume (f/VT).72 The ratio must be calculated dur-
doubles over the course of the trial.7 Before weaning ing spontaneous breathing; calculating it during pres-
is started, however, the respiratory mechanics in such sure support markedly impairs its predictive accura-
patients are similar to those in whom subsequent cy.68 The higher the ratio, the more severe the rapid,
weaning is successful.66 Thus, unknown mechanisms shallow breathing and the greater the likelihood of
associated with the act of spontaneous breathing cause unsuccessful weaning. A ratio of 100 best discrimi-
the worsening of respiratory mechanics in patients nates between successful and unsuccessful attempts
who cannot be weaned from mechanical ventilation. at weaning. In a case of clinical equipoise that is,
In addition to the increase in respiratory effort, an a pretest probability of 50 percent an f/VT of 80,
unsuccessful attempt at spontaneous breathing causes which has a likelihood ratio of 7.5, is associated with
considerable cardiovascular stress.67 Patients can have almost a 95 percent post-test probability of success-
substantial increases in right and left ventricular after- ful weaning.73 If the f/VT is higher than 100, the
load, with increases of 39 and 27 percent in pulmo- likelihood ratio is 0.04 and the post-test probability
nary and systemic arterial pressures, respectively,64 of successful weaning is less than 5 percent.
most likely because the negative swings in intratho- Several groups of investigators have evaluated the
racic pressure are more extreme. At the completion predictive value of f/VT.74-78 Its positive predictive
of a trial of weaning, the level of oxygen consump- value the proportion of patients who are success-
tion is equivalent in patients who can be weaned and fully weaned among those for whom the ratio predicts
in those who cannot. But how the cardiovascular sys- success has generally been high (0.8 or higher).
tem meets the oxygen demand differs in the two The negative predictive value the proportion of
groups of patients.64 In those who are successfully patients who cannot be weaned among those for
weaned, the oxygen demand is met through an in- whom the ratio predicts failure has sometimes
crease in oxygen delivery, mediated by the expected been reported to be low (0.5 or less). Low negative
increase in cardiac output on discontinuation of pos- predictive values have often been reported for pa-
itive-pressure ventilation. In patients who cannot be tients with a high likelihood of successful extubation
weaned, the oxygen demand is met through an in- for example, patients undergoing routine postop-

1992 N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org
 MED IC A L PROGR ES S

Failed Weaning Trial

Mechanical
Ventilation Start End Healthy Subject

Tidal Volume
800
400

(ml)
0
400
0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6
Pleural Pressure
(cm of water)
10
0
10
20
0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6
Pressure (mm Hg)
Pulmonary-Artery

90
60
30
0
0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6
Seconds

Figure 4. Tidal Volume, Pleural Pressure, and Pulmonary-Artery Pressure in a Patient Undergoing Assist-Control Ventilation and at
the Start and End of a Failed Trial of Spontaneous Breathing.
During mechanical ventilation, the patients inspiratory effort is in the normal range and the pulmonary-artery pressure is 45/22
mm Hg (systolic/diastolic). At the start of the trial of spontaneous breathing, the tidal volume falls to 200 ml, the respiratory fre-
quency increases to 33 breaths per minute, there are swings in pleural pressure of 11 cm of water, and the pulmonary-artery pres-
sure at the end of expiration is 60/28 mm Hg. At the end of the trial, 45 minutes later, the tidal volume and respiratory frequency
are unchanged, there are swings in pleural pressure of 19 cm of water, autopositive end-expiratory pressure is 4 cm of water, and
the pulmonary-artery pressure is 60/31 mm Hg. The values in a healthy subject are tidal volume, 380 ml; respiratory frequency, 17
breaths per minute; pleural-pressure swings, 3 cm of water; and pulmonary-artery pressure, 18/8 mm Hg. Data are from Tobin et
al.63,64 and Jubran et al.7,65

erative ventilatory assistance and patients who have been well defined. For example, pressure support of
tolerated initial trials of weaning.75,76 6 to 8 cm of water is widely used to compensate for
There are four methods of weaning.79 The oldest the resistance imposed by the endotracheal tube and
method is to perform trials of spontaneous breath- ventilator circuit.80 A patient who can breathe com-
ing several times a day, with the use of a T-tube circuit fortably at this level of pressure support should be
containing an enriched supply of oxygen. Initially 5 to able to tolerate extubation. But if the upper airways
10 minutes in duration, the trials are extended and are swollen because an endotracheal tube has been in
repeated several times a day until the patient can sus- place for several days, the work engendered by breath-
tain spontaneous ventilation for several hours. This ing through the swollen airways is about the same as
approach has become unpopular because it requires that caused by breathing through an endotracheal
considerable time on the part of intensive care staff. tube.81 Accordingly, any amount of pressure support
The two most common approaches, intermittent overcompensates and may give misleading informa-
mandatory ventilation and pressure support, decrease tion about the likelihood that a patient can tolerate
ventilatory assistance gradually by respectively low- extubation.
ering the number of ventilator-assisted breaths or the The fourth method of weaning is to perform a
level of pressure. When a minimal level of ventilatory single daily T-tube trial, lasting for up to two hours.
assistance can be tolerated, the patient is extubated. If this trial is successful, the patient is extubated; if
The minimal level of assistance, however, has never the trial is unsuccessful, the patient is given at least

N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org 1993
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

24 hours of respiratory-muscle rest with full ventila- ventilation16; they have not yet been approved for gen-
tory support before another trial is performed.82 eral clinical use.
Until the early 1990s, it was widely believed that
all weaning methods were equally effective, and the CONCLUSIONS
physicians judgment was regarded as the critical de- Since my previous overview of mechanical ventila-
terminant. But the results of randomized, controlled tion in the Journal, we have gained a better under-
trials clearly indicate that the period of weaning is as standing of the pathophysiology associated with un-
much as three times as long with intermittent man- successful weaning and have learned how to wean
datory ventilation as with trials of spontaneous patients more efficiently. We have also learned how
breathing.61,62 In a study involving patients with res- ventilator settings influence survival in patients with
piratory difficulties on weaning, trials of spontane- the acute respiratory distress syndrome. Less progress
ous breathing halved the weaning time as compared has been made in determining how the ventilator can
with pressure support62; in another study, the weaning best be used to achieve maximal respiratory-muscle
time was similar with the two methods.61 Perform- rest, which is the most common reason for providing
ing trials of spontaneous breathing once a day is mechanical ventilation. Although further research may
as effective as performing such trials several times a lead to unexpected new insights, an important chal-
day62 but much simpler. In a recent study, half-hour lenge for researchers is to identify elements of our
trials of spontaneous breathing were as effective as current knowledge that can be incorporated into a
two-hour trials.83 However, this study involved all pa- clinical management scheme to improve the outcome
tients being considered for weaning, not just those for patients who require ventilatory assistance.
for whom there were difficulties with weaning.
A two-stage approach to weaning systematic Supported by a Merit Review grant from the Department of Veterans
measurement of predictors, including f/VT , followed Affairs Research and Development Service.
by a single daily trial of spontaneous breathing was
compared with conventional management in a ran- I am indebted to Drs. Amal Jubran, Franco Laghi, and Thomas
Brack for helpful criticisms on successive drafts of the manuscript.
domized trial.69 Although the patients assigned to the
two-stage approach were sicker than those assigned REFERENCES
to conventional weaning, they were weaned twice as
rapidly. The rate of complications and the costs of 1. Tobin MJ. Mechanical ventilation. N Engl J Med 1994;330:1056-61.
2. Esteban A, Anzueto A, Alia I, et al. How is mechanical ventilation em-
intensive care were also lower with two-stage man- ployed in the intensive care unit? An international utilization review. Am J
agement than with conventional management. Respir Crit Care Med 2000;161:1450-8.
When patients can sustain spontaneous ventilation 3. Sassoon CSH. Intermittent mandatory ventilation. In: Tobin JM, ed.
Principles and practice of mechanical ventilation. New York: McGraw-Hill,
without undue discomfort, they are extubated. About 1994:221-37.
10 to 20 percent of such patients require reintuba- 4. Marini JJ, Smith TC, Lamb VJ. External work output and force gener-
tion.61,62 Mortality among patients who require re- ation during synchronized intermittent mechanical ventilation: effect of
machine assistance on breathing effort. Am Rev Respir Dis 1988;138:
intubation is more than six times as high as mortality 1169-79.
among patients who can tolerate extubation.83,84 The 5. Brochard L. Pressure support ventilation. In: Tobin MJ, ed. Principles
and practice of mechanical ventilation. New York: McGraw-Hill, 1994:
reason for the higher mortality is unknown; it is not 239-57.
clearly related to the development of new problems 6. Jubran A, Van de Graaff WB, Tobin MJ. Variability of patient-ventilator
after extubation or to complications of reinserting interaction with pressure support ventilation in patients with chronic ob-
structive pulmonary disease. Am J Respir Crit Care Med 1995;152:129-36.
the tube. Indeed, the need for reintubation may sim- 7. Jubran A, Tobin JM. Pathophysiologic basis of acute respiratory distress
ply be a marker of a more severe underlying illness. in patients who fail a trial of weaning from mechanical ventilation. Am J
In a controlled trial involving patients who could Respir Crit Care Med 1997;155:906-15.
8. Reid WD, Huang J, Bryson S, Walker DC, Belcastro AN. Diaphragm
not sustain spontaneous ventilation, the patients who injury and myofibrillar structure induced by resistive loading. J Appl Phys-
were extubated and then received noninvasive venti- iol 1994;76:176-84.
9. Leung P, Jubran A, Tobin MJ. Comparison of assisted ventilator modes
lation through a face mask had a shorter mean overall on triggering, patient effort, and dyspnea. Am J Respir Crit Care Med
period of ventilatory support (10.2 days) than those 1997;155:1940-8.
who remained intubated and were weaned by decreas- 10. Anzueto A, Peters JI, Tobin JM, et al. Effects of prolonged controlled
mechanical ventilation on diaphragmatic function in healthy adult ba-
ing pressure support (16.6 days).85 Although this re- boons. Crit Care Med 1997;25:1187-90.
sult is promising, it is not clear how many such pa- 11. Parthasarathy S, Jubran A, Tobin MJ. Cycling of inspiratory and expir-
tients or which ones could benefit from this approach. atory muscle groups with the ventilator in airflow limitation. Am J Respir
Crit Care Med 1998;158:1471-8. [Erratum, Am J Respir Crit Care Med
1999;159:1023.]
OTHER APPROACHES TO MECHANICAL 12. Simon PM, Zurob AS, Wies WM, Leiter JC, Hubmayr RD. Entrain-
VENTILATION ment of respiration in humans by periodic lung inflations: effect of state
and CO2. Am J Respir Crit Care Med 1999;160:950-60.
Noninvasive ventilation, an approach that is be- 13. Parthasarathy S, Jubran A, Tobin MJ. Assessment of neural inspiratory
coming more widespread, was reviewed in the Jour- time in ventilator-supported patients. Am J Respir Crit Care Med 2000;
162:546-52.
nal in 1997.86 Two new approaches under investiga- 14. Sassoon CSH, Gruer SE. Characteristics of the ventilator pressure- and
tion are liquid ventilation87 and proportional-assist flow-trigger variables. Intensive Care Med 1995;21:159-68.

1994 N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org
 MED IC A L PROGR ES S

15. Marini JJ, Capps JS, Culver BH. The inspiratory work of breathing 42. Vieira SR, Puybasset L, Lu Q, et al. A scanographic assessment of pul-
during assisted mechanical ventilation. Chest 1985;87:612-8. monary morphology in acute lung injury: significance of the lower inflec-
16. Giannouli E, Webster K, Roberts D, Younes M. Response of ventila- tion point detected on the lung pressure-volume curve. Am J Respir Crit
tor-dependent patients to different levels of pressure support and propor- Care Med 1999;159:1612-23.
tional assist. Am J Respir Crit Care Med 1999;159:1716-25. 43. Vieira SR, Puybasset L, Richecoeur J, et al. A lung computed tomo-
17. Sinderby C, Navalesi P, Beck J, et al. Neural control of mechanical ven- graphic assessment of positive end-expiratory pressure-induced lung over-
tilation in respiratory failure. Nat Med 1999;5:1433-6. distension. Am J Respir Crit Care Med 1998;158:1571-7.
18. Puddy A, Younes M. Effect of inspiratory flow rate on respiratory out- 44. Horton WG, Cheney FW. Variability of effect of positive end expira-
put in normal subjects. Am Rev Respir Dis 1992;146:787-9. tory pressure. Arch Surg 1975;110:395-8.
19. Laghi F, Karamchandani K, Tobin MJ. Influence of ventilator settings 45. Kanarek DJ, Shannon DC. Adverse effect of positive end-expiratory
in determining respiratory frequency during mechanical ventilation. Am J pressure on pulmonary perfusion and arterial oxygenation. Am Rev Respir
Respir Crit Care Med 1999;160:1766-70. Dis 1975;112:457-9.
20. Pierson DJ. Barotrauma and bronchopleural fistula. In: Tobin MJ, ed. 46. Gattinoni L, Pelosi P, Suter PM, Pedoto A, Vercesi P, Lissoni A. Acute
Principles and practice of mechanical ventilation. New York: McGraw-Hill, respiratory distress syndrome caused by pulmonary and extrapulmonary
1994:813-36. disease: different syndromes? Am J Respir Crit Care Med 1998;158:3-11.
21. Webb HH, Tierney DF. Experimental pulmonary edema due to inter- 47. Lu Q, Vieira SR, Richecoeur J, et al. A simple automated method for
mittent positive pressure ventilation with high inflation pressures: protec- measuring pressure-volume curves during mechanical ventilation. Am J
tion by positive end-expiratory pressure. Am Rev Respir Dis 1974;110: Respir Crit Care Med 1999;159:275-82.
556-65. 48. Argiras EP, Blakeley CR, Dunnill MS, Otremski S, Sykes MK. High
22. Dreyfuss D, Saumon G. Ventilator-induced lung injury: lessons from PEEP decreases hyaline membrane formation in surfactant deficient lungs.
experimental studies. Am J Respir Crit Care Med 1998;157:294-323. Br J Anaesth 1987;59:1278-85.
23. Puybasset L, Cluzel P, Chao N, Slutsky AS, Coriat P, Rouby JJ. 49. Tremblay L, Valenza F, Ribeiro SP, Li J, Slutsky AS. Injurious ventila-
A computed tomography scan assessment of regional lung volume in acute tory strategies increase cytokines and c-fos m-RNA expression in an isolat-
lung injury. Am J Respir Crit Care Med 1998;158:1644-55. ed rat lung model. J Clin Invest 1997;99:944-52.
24. Gattinoni L, Pesenti A, Bombino M, et al. Relationships between lung 50. Mead J, Takishima T, Leith D. Stress distribution in lungs: a model of
computed tomographic density, gas exchange, and PEEP in acute respira- pulmonary elasticity. J Appl Physiol 1970;28:596-608.
tory failure. Anesthesiology 1988;69:824-32. 51. Hickling KG. The pressure-volume curve is greatly modified by re-
25. Rouby JJ, Lherm T, Martin de Lassale E, et al. Histologic aspects of cruitment: a mathematical model of ARDS lungs. Am J Respir Crit Care
pulmonary barotrauma in critically ill patients with acute respiratory fail- Med 1998;158:194-202.
ure. Intensive Care Med 1993;19:383-9. 52. Martynowicz MA, Minor TA, Walters BJ, Hubmayr RD. Regional ex-
26. Hickling KG, Henderson SJ, Jackson R. Low mortality associated with pansion of oleic acid-injured lungs. Am J Respir Crit Care Med 1999;160:
low volume pressure limited ventilation with permissive hypercapnia in se- 250-8.
vere adult respiratory distress syndrome. Intensive Care Med 1990;16:372- 53. Lichtwarck-Aschoff M, Mols G, Hedlund AJ, et al. Compliance is non-
7. linear over tidal volume irrespective of positive end-expiratory pressure lev-
27. Amato MB, Barbas CS, Medeiros DM, et al. Beneficial effects of the el in surfactant-depleted piglets. Am J Respir Crit Care Med 2000;162:
open lung approach with low distending pressures in acute respiratory 2125-33.
distress syndrome: a prospective randomized study on mechanical ventila- 54. Carney DE, Bredenberg CE, Schiller HJ, et al. The mechanism of
tion. Am J Respir Crit Care Med 1995;152:1835-46. lung volume change during mechanical ventilation. Am J Respir Crit Care
28. Amato MBP, Barbas CSV, Medeiros DM, et al. Effect of a protective- Med 1999;160:1697-702.
ventilation strategy on mortality in the acute respiratory distress syndrome. 55. Harris RS, Hess DR, Venegas JG. An objective analysis of the pres-
N Engl J Med 1998;338:347-54. sure-volume curve in the acute respiratory distress syndrome. Am J Respir
29. The Acute Respiratory Distress Syndrome Network. Ventilation with Crit Care Med 2000;161:432-9.
lower tidal volumes as compared with traditional tidal volumes for acute 56. Pelosi P, Tubiolo D, Mascheroni D, et al. Effects of the prone position
lung injury and the acute respiratory distress syndrome. N Engl J Med on respiratory mechanics and gas exchange during acute lung injury. Am
2000;342:1301-8. J Respir Crit Care Med 1998;157:387-93.
30. Stewart TE, Meade MO, Cook DJ, et al. Evaluation of a ventilation 57. Gattinoni L, Pelosi P, Vitale G, Pesenti A, DAndrea L, Mascheroni D.
strategy to prevent barotrauma in patients at high risk for acute respiratory Body position changes redistribute lung computed-tomographic density in
distress syndrome. N Engl J Med 1998;338:355-61. patients with acute respiratory failure. Anesthesiology 1991;74:15-23.
31. Brochard L, Roudot-Thoraval F, Roupie E, et al. Tidal volume reduc- 58. Mutoh T, Guest RJ, Lamm WJ, Albert RK. Prone position alters the
tion for prevention of ventilator-induced lung injury in acute respiratory effect of volume overload on regional pleural pressures and improves hy-
distress syndrome. Am J Respir Crit Care Med 1998;158:1831-8. poxemia in pigs in vivo. Am Rev Respir Dis 1992;146:300-6.
32. Brower RG, Shanholtz CB, Fessler HE, et al. Prospective, random- 59. Mure M, Domino KB, Lindahl SG, Hlastala MP, Altemeier WA, Glenny
ized, controlled clinical trial comparing traditional versus reduced tidal vol- RW. Regional ventilation-perfusion distribution is more uniform in the
ume ventilation in acute respiratory distress syndrome patients. Crit Care prone position. J Appl Physiol 2000;88:1076-83.
Med 1999;27:1492-8. 60. Albert RK, Hubmayr RD. The prone position eliminates compression
33. Tobin MJ. Culmination of an era in research on the acute respiratory of the lungs by the heart. Am J Respir Crit Care Med 2000;161:1660-5.
distress syndrome. N Engl J Med 2000;342:1360-1. 61. Brochard L, Rauss A, Benito S, et al. Comparison of three methods
34. Brochard L. Respiratory pressure-volume curves. In: Tobin MJ, ed. of gradual withdrawing from ventilatory support during weaning from me-
Principles and practice of intensive care monitoring. New York: McGraw- chanical ventilation. Am J Respir Crit Care Med 1994;150:896-903.
Hill, 1998:597-616. 62. Esteban A, Frutos F, Tobin MJ, et al. A comparison of four methods
35. Roupie E, Dambrosio M, Servillo G, et al. Titration of tidal volume of weaning patients from mechanical ventilation. N Engl J Med 1995;332:
and induced hypercapnia in acute respiratory distress syndrome. Am J Res- 345-50.
pir Crit Care Med 1995;152:121-8. 63. Tobin MJ, Perez W, Guenther SM, et al. The pattern of breathing dur-
36. Richecoeur J, Lu Q, Vieira SR, et al. Expiratory washout versus opti- ing successful and unsuccessful trials of weaning from mechanical ventila-
mization of mechanical ventilation during permissive hypercapnia in pa- tion. Am Rev Respir Dis 1986;134:1111-8.
tients with severe acute respiratory distress syndrome. Am J Respir Crit 64. Tobin MJ, Chadha TS, Jenouri G, Birch SJ, Gazeroglu HB, Sackner
Care Med 1999;160:77-85. MA. Breaking patterns. 1. Normal subjects. Chest 1983;84:202-5.
37. Feihl F, Perret C. Permissive hypercapnia: how permissive should we 65. Jubran A, Mathru M, Dries D, Tobin MJ. Continuous recordings of
be? Am J Respir Crit Care Med 1994;150:1722-37. mixed venous oxygen saturation during weaning from mechanical ventila-
38. Feihl F, Eckert P, Brimioulle S, et al. Permissive hypercapnia impairs tion and the ramifications thereof. Am J Respir Crit Care Med 1998;158:
pulmonary gas exchange in the acute respiratory distress syndrome. Am J 1763-9.
Respir Crit Care Med 2000;162:209-15. 66. Jubran A, Tobin MJ. Passive mechanics of lung and chest wall in pa-
39. Mead J, Collier C. Relation of volume history of lungs to respiratory tients who failed or succeeded in trials of weaning. Am J Respir Crit Care
mechanics in anesthetized dogs. J Appl Physiol 1959;14:669-78. Med 1997;155:916-21.
40. Pelosi P, Cadringher P, Bottino N, et al. Sigh in acute respiratory dis- 67. Lemaire F, Teboul JL, Cinotti L, et al. Acute left ventricular dysfunc-
tress syndrome. Am J Respir Crit Care Med 1999;159:872-80. tion during unsuccessful weaning from mechanical ventilation. Anesthesi-
41. Dambrosio M, Roupie E, Mollett JJ, et al. Effects of positive end- ology 1988;69:171-9.
expiratory pressure and different tidal volumes on alveolar recruitment and 68. Stroetz RW, Hubmayr RD. Tidal volume maintenance during weaning
hyperinflation. Anesthesiology 1997;87:495-503. with pressure support. Am J Respir Crit Care Med 1995;152:1034-40.

N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org 1995
 The Ne w E n g l a nd Jo u r n a l o f Me d ic i ne

69. Ely EW, Baker AM, Dunagan DP, et al. Effect on the duration of me- gas tonometry and weaning from mechanical ventilation. Am J Respir Crit
chanical ventilation of identifying patients capable of breathing spontane- Care Med 2000;161:171-6.
ously. N Engl J Med 1996;335:1864-9. 79. Tobin MJ. 1999 Donald F Egan Scientific Lecture: weaning from me-
70. Coplin WM, Pierson DJ, Cooley KD, Newell DW, Rubenfeld GD. Im- chanical ventilation: what have we learned? Respir Care 2000;45:417-31.
plications of extubation delay in brain-injured patients meeting standard 80. Brochard L, Rua F, Lorino H, Lemaire F, Harf A. Inspiratory pressure
weaning criteria. Am J Respir Crit Care Med 2000;161:1530-6. support compensates for the additional work of breathing caused by the
71. Tobin MJ, Alex CG. Discontinuation of mechanical ventilation. In: To- endotracheal tube. Anesthesiology 1991;75:739-45.
bin MJ, ed. Principles and practice of mechanical ventilation. New York: 81. Straus C, Louis B, Isabey D, Lemaire F, Harf A, Brochard L. Contri-
McGraw-Hill, 1994:1177-206. bution of the endotracheal tube and the upper airway to breathing work-
72. Yang KL, Tobin MJ. A prospective study of indexes predicting the out- load. Am J Respir Crit Care Med 1998;157:23-30.
come of trials of weaning from mechanical ventilation. N Engl J Med 1991; 82. Laghi F, DAlfonso N, Tobin MJ. Pattern of recovery from diaphrag-
324:1445-50. matic fatigue over 24 hours. J Appl Physiol 1995;79:539-46.
73. Jaeschke RZ, Meade MO, Guyatt GH, Keenan SP, Cook DJ. How to 83. Esteban A, Alia I, Tobin MJ, et al. Effect of spontaneous breathing
use diagnostic test articles in the intensive care unit: diagnosing weanability trial duration on outcome of attempts to discontinue mechanical ventila-
using f/Vt. Crit Care Med 1997;25:1514-21. tion. Am J Respir Crit Care Med 1999;159:512-8.
74. Sassoon CSH, Mahutte CK. Airway occlusion pressure and breathing 84. Epstein SK, Ciubotaru RL, Wong JB. Effect of failed extubation on
pattern as predictors of weaning outcome. Am Rev Respir Dis 1993;148: the outcome of mechanical ventilation. Chest 1997;112:186-92.
860-6. 85. Nava S, Ambrosino N, Clini E, et al. Noninvasive mechanical ventila-
75. Epstein SK. Etiology of extubation failure and the predictive value of tion in the weaning of patients with respiratory failure due to chronic ob-
the rapid shallow breathing index. Am J Respir Crit Care Med 1995;152: structive pulmonary disease: a randomized, controlled trial. Ann Intern
545-9. Med 1998;128:721-8.
76. Jacob B, Chatila W, Manthous CA. The unassisted respiratory rate/ 86. Hillberg RE, Johnson DC. Noninvasive ventilation. N Engl J Med
tidal volume ratio accurately predicts weaning outcome in postoperative 1997;337:1746-52.
patients. Crit Care Med 1997;25:253-7. 87. Leach CL, Greenspan JS, Rubenstein SD, et al. Partial liquid ventila-
77. Vallverdu I, Calaf N, Subirana M, Net A, Benito S, Mancebo J. Clinical tion with perflubron in premature infants with severe respiratory distress
characteristics, respiratory functional parameters, and outcome of a two- syndrome. N Engl J Med 1996;335:761-7.
hour T-piece trial in patients weaning from mechanical ventilation. Am J
Respir Crit Care Med 1998;158:1855-62.
78. Maldonado A, Bauer TT, Ferrer M, et al. Capnometric recirculation Copyright 2001 Massachusetts Medical Society.

ELECTRONIC ACCESS TO THE JOURNAL S CUMULATIVE INDEX

At the Journals site on the World Wide Web (http://www.nejm.org) you can search an
index of all articles published since January 1975 (abstracts 1975-1992, full-text 1993-
present). You can search by author, key word, title, type of article, and date. The results will
include the citations for the articles plus links to the abstracts of articles published since
1993. Time-limited access to single articles and 24-hour site access can also be ordered for
a fee through the Internet (http://www.nejm.org).

1996 N Engl J Med, Vol. 344, No. 26 June 28, 2001 www.nejm.org