High-frequency mechanical ventilation
Principles and practices in the era
of lung-protective ventilation strategies
Jeff M. Singh, MDa, Thomas E. Stewart, MDa,b,*
aDepartment of Medicine and the Interdepartmental Division of Critical Care Medicine,
University of Toronto, Toronto, Ontario, M5G 1X5, Canada
bIntensive Care Unit, Mount Sinai Hospital, 600 University Avenue, Toronto, Ontario,
M5G 1X5, Canada
Mechanical ventilation is the cornerstone of care for patients in respira-
tory failure. Conventional positive-pressure mechanical ventilation (CMV)
at near-physiologic tidal volumes has become the most popular form
because of its simplicity and widespread availability. Recent evidence, how-
ever, suggests that mechanical ventilation itself both potentiates and causes
lung injury in patients with severe respiratory disease. This observation has
prompted a search for safer ventilation strategies that provide optimal ven-
tilatory support while minimizing the potential for causing lung injury.
The term high-frequency ventilation (HFV) refers to mechanical ventila-
tion modes that use high respiratory rates and small tidal volumes-often
smaller than anatomic dead space. First developed in the 1960s, HFV was
the subject of considerable research. Despite theoretical advantages that
support the use of HFV in patients with injured lungs, initial clinical studies
failed to demonstrate a significant advantage over CMV [1–3]. Subsequently,
interest in HFV waned.
Recently, there has been renewed interest in some forms of HFV, particu-
larly high-frequency oscillatory ventilation (HFO). This renewed interest has
been fueled by the belief that HFO may be well suited for a lung-protective
ventilation strategy. Early clinical studies that have evaluated HFO in this
regard have been promising.
This review provides an overview of HFV, its mechanical properties, and
the rationale that supports its use in this era of lung-protective mechanical
ventilation.
Respir Care Clin 8 (2002) 247–260
* Corresponding author.
E-mail address: tom.stewart@utoronto.ca (T.E. Stewart).
1078-5337/02/$ - see front matter � 2002, Elsevier Science (USA). All rights reserved.
PII: S 1 0 7 8 - 5 3 3 7 ( 0 2 ) 0 0 0 0 7 - 2
Principles
Before the 1950s, negative-pressure ventilation was the ventilation mode
of choice. Not until the polio epidemic of the 1950s did a lack of negative-
pressure ventilators (so-called ‘‘iron lungs’’) prompt the development of
positive-pressure mechanical ventilators. These new ventilators quickly be-
came the standard of care for patients in respiratory failure. In the 40 years
that followed, positive-pressure ventilation was applied typically with a
volume- or pressure-cycled mode and often used high-driving pressures to
inflate lungs to physiologic or supraphysiologic volumes.
The observation that sufficient ventilation in dogs occurred with tidal
volumes of less than anatomic dead space was made in 1915 [4]. Over 50
years passed, however, before this concept was applied to mechanical venti-
lation in humans [5]. Nonetheless, the original observation did stimulate
research into the physiology of gas exchange using high frequencies and low
tidal volumes.
Theories of gas transport
The adequate ventilation that was achieved by using small volumes in
animal models implied that gas exchange during positive-pressure ventila-
tion is complex and may be only partially dependent on bulk flow. There are
physiologic data that support the role of several mechanisms of gas trans-
port during HFV.
Bulk flow
If a sufficiently large volume of gas is inspired, the leading edge of the gas
front may reach the proximal airways and a few proximal alveoli [6]. At
tidal volumes greater than one-half of the anatomic dead space, this may
be a contributing mechanism of gas transport and may account for the sud-
den decrease in ventilation efficiency in high-frequency modes when the tidal
volumes decrease below this volume [7].
Pendelluft
The rate at which a lung unit is filled with fresh gas when inflated by a
given driving pressure is determined by its mechanical properties, as
expressed by its time constant. Heterogeneity of time constants between
lung units has long been recognized [8]. Because of this heterogeneity, some
lung units experience more rapid filling than others during a given inspira-
tion. After inspiration, gas is redistributed from fast- to slow-filling units, with
redistribution in the opposite direction at end-expiration. This mechanism
may enhance alveolar gas exchange, particularly of peripheral lung units,
and may help gas mixing within the airways [9].
248 J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
Asymmetric velocity profiles
Schroter and Sudlow [10] first demonstrated that inspiratory velocity pro-
files in bifurcating systems were more skewed during inspiration than during
expiration [10]. After repeated cycles of inspiration and expiration, a central
current delivers gas distally and the gas at the periphery of the airways
returns proximally.
Taylor dispersion
The aforementioned asymmetric velocity profile of inspired air creates a
radial concentration gradient in the airway and, thus, turbulent eddies.
These eddies, which augment the mixing of gas in the airways, are produced
by the combined effect of convective flow and diffusion [11,12].
Cardiogenic mixing and diffusion
As in conventional modes of ventilation, the mechanical agitation of lung
units adjacent to the beating heart and the molecular diffusion along concen-
tration gradients near the alveolocapillary membrane may play important
roles in gas transport. The mechanisms of gas exchange during HFV are
undoubtedly complex, and many or all of the aforementioned mechanisms
may contribute to gas exchange in a patient at any given moment. Perhaps
most important to the clinician is that adequate ventilation can be achieved
with judicious selection of patient and ventilation mode, despite the use of
very low tidal volumes.
Modes of high-frequency ventilation
As HFV has advanced, different classes of this type of ventilation have
emerged. They differ in the mechanism with which driving pressure and,
hence, tidal volume are generated. HFV also can be classified according
to the exhalation phase; most classes of HFV depend on the passive recoil
of the respiratory system for exhalation. HFO, however, develops a negative
pressure gradient to create an active expiration phase. The unique character-
istics, limitations, and potential complications of some of the most popular
modes of HFV are outlined in the following sections.
Passive exhalation
High-frequency jet ventilation (HFJV) was developed in 1967 and sub-
sequently gained some popularity. HFJV delivers tidal volumes of 2–5 cm3/kg
at a rate of 100 to 200 breaths per minute. A system of valves controls the gas
stream, which is delivered at high pressure through a narrow-bore cannula
into the proximal endotracheal tube and is conditioned with aerosolized
saline. The small-bore gas cannula reduces visual obstruction in the airway,
a characteristic that has led to HFJV’s popularity in rigid bronchoscopy.
Because the high-velocity jet entrains air from the circuit, precisely control-
ling tidal volume is difficult. Additionally, gas warming and humidification
249J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
can be problematic in HFJV because the gas jet both expands and cools as it
exits the high-pressure cannula. Suboptimal gas conditioning and high gas-
flow rates have been implicated in the damage to the trachea and proximal
airways that has been observed with HFJV [13–17]. Expiration in HFJV is
passive and depends on the elastic recoil of the respiratory system. Thus, at
high respiratory frequencies, the clinician must be wary of gas trapping with
its associated lung injury and hemodynamic complications.
High-frequency positive-pressure ventilation was developed to provide
adequate mechanical ventilation in circumstances during which minimal
movement of the upper airway is desired, such as laryngeal surgery. Small
tidal volumes of gas are delivered at rates between 60 and 100 breaths per
minute. As with HFJV, expiration is passive, and the risk of gas trapping
increases as the respiratory time decreases at higher respiratory rates.
Active exhalation
First reported in a letter in 1972 by Lunkenheimer et al. [18,19], HFO uti-
lizes a reciprocating diaphragm to create oscillations in the air column. This
generates a driving pressure that results in gas exchange and ventilation
(Fig. 1). As the diaphragm drives pressure waves in both directions, inspira-
tion and expiration are active. This may reduce the potential for gas trap-
ping, particularly at high mean airway pressures, thereby allowing for
tighter control of lung volumes.
In HFO, mean airway pressure is determined by controlling the flow of
fresh gas through the circuit, known as ‘‘bias flow.’’ The result is the effective
uncoupling of oxygenation and ventilation, whereby (1) mean airway pres-
sure is adjusted to control lung volume and oxygenation, and (2) the rate
Fig. 1. Schematic diagram of an HFO ventilator and circuit (Adapted fromHess D, Mason BR,
High-frequency ventilation: design and equipment issues. Respir Care Clin North Am 2001;
7:577–98.)
250 J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
and the displacement of the oscillating diaphragm are varied to achieve ade-
quate ventilation. During HFO, the airway pressures are thought to oscillate
at pressures close to the mean airway pressure (Fig. 2), which avoids large
pressure swings and therefore may help reduce iatrogenic lung injury.
Additionally, HFO provides considerable advantages over other forms of
HFV in gas conditioning. The lack of gas entrainment and the strict control
of bias flow in HFO permit optimal warming and humidification of inspired
gas. Tracheal damage similar to that seen in HFJV, however, has been re-
ported and must be monitored closely [13,15].
Rationale
Recently, a lung-protective ventilation strategy designed to minimize ven-
tilator-induced lung injury (VILI), was demonstrated to reduce mortality in
patients with acute respiratory distress syndrome (ARDS) [20,21]. The pro-
ven benefits of lung-protective ventilation have spurred renewed interest
in HFV, particularly HFO. The advantages of HFO with respect to volume
control and gas conditioning may make it well suited to lung-protective
mechanical ventilation in an adult population. Understanding the rationale
for using HFO in such a strategy requires some knowledge of the pathogen-
esis of VILI, which is reviewed briefly in the following section.
Current concepts of ventilator-induced lung injury
Respiratory mechanics are altered significantly in injured or diseased
lungs. Respiratory-system compliance often is decreased, necessitating higher
driving pressures to maintain adequate tidal volumes and ventilation. More-
over, disease pathology is not distributed uniformly throughout the lung.
Consequently, there is regional heterogeneity in lung compliance [22,23]. App-
lying positive-pressure ventilation in this scenario can result in overinflation
Fig. 2. Diagram that compares pressure oscillations throughout the respiratory cycle between
conventional mechanical ventilation (CMV) and high-frequency oscillatory ventilation (HFO).
(Adapted from Ferguson ND, Stewart T. Use of high-frequency oscillatory ventilation in adults
with acute lung injury. Respir Care Clin North Am 2002;18:98.)
251J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
of more healthy lung units while injuring areas of lung collapse. Animals
that are subjected to mechanical ventilation at extreme airway pressures
and volumes demonstrate severe alterations in lung permeability and
histologic damage similar to that observed in human ARDS patients
[24–26].
Although the exact pathogenesis of VILI remains controversial, several
contributory mechanisms have been elucidated. Current models of injury
include volutrauma, atelectrauma, and oxygen toxicity. Recently, these
mechanical and chemical insults have been reported to activate inflamma-
tory cascades, resulting in further lung injury and contributing to multisys-
tem organ failure. This process has been termed biotrauma.
The static pressure-volume curve (P-V curve) of the total respiratory sys-
tem may be useful to help understand concepts of VILI. Within the P-V
curve, two points often can be identified: the lower inflection point (LIP) and
the upper inflection point (UIP) (Fig. 3). Some have argued that LIP may
represent the point in inspiration when lung units begin to open (recruit-
ment); hence, some authors have advocated setting PEEP above this level
to avoid atelectasis [21,27]. TheLIP represents the pressure atwhichmost lung
units open, not the pressure at which all lung units open. Considerable het-
erogeneity may exist in lung-opening pressures. Recent animal and human
data have demonstrated that lung recruitment in injured lungs (ARDS) is
progressive and continuous over the P-V curve regardless of inflection
points, possibly clouding the clinical use of LIP measurements [28]. The UIP
may represent the point at which most of the lung is overdistended and
prone to injury; thus, some have suggested that peak pressure should be
maintained below this level to reduce the potentially deleterious effects [29].
Fig. 3. An example of a P-V curve that depicts the total respiratory system in an ARDS patient.
Note the lower and upper inflection points on the inspiratory limb and the zones of potential
atelectrauma and volutrauma.
252 J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
Volutrauma
High-volume ventilation may overdistend both normal and diseased
alveoli and thereby produce lung injury. Some have suggested that ventilator-
induced pulmonary edema may not be caused by high airway pressures but
by the large lung volumes with which they are associated. One animal study
found that rats ventilated with high tidal volumes and high pressures rapidly
developed pulmonary edema and ultrastructural abnormalities. When the
rats were ventilated with high pressures but with tidal volumes that were
limited by mechanical strapping, they did not develop pulmonary edema
[24]. Moreover, pulmonary edema has been observed to develop rapidly
in rats ventilated to high tidal volumes by negative-pressure mechanical ven-
tilation [24]. Because these findings have been reproduced in other animal
models, [30,31] the accepted theory has become that alveolar distention
from large ventilatory volumes, not the associated high airway pressure, is
the key contributor to ventilator-induced pulmonary edema. Accordingly,
the term barotrauma has been replaced with the term volutrauma.
To reduce the risk of regional lung-unit overdistention, some have sug-
gested that lung-protective CMV should incorporate low tidal volumes
(6–8 mL/kg) and moderate peak airway pressures. Recently, the clinical ben-
efit of lower-tidal-volume ventilation in the context of ARDS was supported
by a large multicenter trial by the National Institutes of Health ARDS net-
work [20]. In this landmark study, a lung-protective strategy of lower tidal
volumes (<6 mL/kg) and plateau pressures (<30 cm H2O) was compared in
a prospective, randomized fashion with a higher-tidal-volume strategy. The
observed reduction in absolute mortality was 9%.
Aside from the benefit of low-tidal-volume CMV, there may be theoreti-
cal advantages to HFV. One study found that 80% of ARDS patients who
were ventilated with CMV had plateau pressures above that which corre-
sponded to UIP on their P-V curves. As mentioned previously, this intro-
duced the possibility of the overdistention of compliant lung units [29].
Although this study used tidal volumes larger than those that have been pro-
ven beneficial in ARDS [20], plateau pressures were maintained below UIP
only at small tidal volumes (5.5 mL/kg). These observations imply that even
lung-protective strategies based on CMV may cause volutrauma-type injury
to the lung. The theoretical advantage to the HFV modes lies in the small
tidal volumes and small pressure swings, both of which reduce the risk of
regional lung-unit distention and volutrauma.
Atelectrauma
Animal studies have demonstrated that ventilation at low lung volumes
induces injury to the alveolar wall because of the shear stresses generated
by the repeated opening and closing of alveoli during the respiratory cycle
[32]. This shearing injury occurs when atelectatic lung regions are forced
open repetitively and then allowed to collapse on expiration. If expiratory
253J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
volumes are low, lung units collapse at end-expiration and set the stage for
atelectrauma.
A recent animal study demonstrated that healthy lungs tolerated these
injurious forces for short periods without histologic damage [33]. In the con-
text of surfactant dysfunction or deficiency similar to that seen in ARDS,
however, severe histologic damage was observed after only brief ventilation
[34,35]. In an attempt to reduce the potential damage from such injury, some
clinicians have advocated the use of PEEP to maintain pressures above the
alveolar closing pressure. This strategy is thought to maintain lung volume
and to prevent lung-unit collapse at end-exhalation. One study that targeted
PEEP levels above LIP on the static P-V curve demonstrated increased sur-
vival at 28 days; however, this was a relatively small study (n¼ 53) and had
a mortality rate in the control group that was higher than normally expected
[21]. Despite data indicating that the application of high PEEP level is ben-
eficial, the optimal level of PEEP for reducing ventilator-associated lung
injury remains elusive.
Oxygen toxicity
Exposure to high concentrations of fractional inspired oxygen was one of
the first mechanisms to be linked to VILI. This phenomenon has been well
studied and is accepted widely. In addition to VILI, oxygen toxicity may
potentiate lung injury through absorption atelectasis, secondary hypoventi-
lation, and systemic vasoconstriction [36]. Despite these observations, the
exact oxygen level that results in toxicity has not been determined. Similarly,
the relative contribution of volutrauma, atelectrauma, and oxygen toxicity
to lung damage are unknown.
Biotrauma
Mortality in ARDS has been reported to be between 35% to 65%. Most
deaths, however, do not result from respiratory failure but from progressive
multiple-organ dysfunction [37]. Inflammatory mediators released in the
lungs because of primary lung disease or secondary injury may contribute
to widespread systemic inflammation and multiple-organ dysfunction
[38,39]. Several early trials have observed that lung-protective ventilation
strategies are associated with lower levels of systemic inflammatory media-
tors [20,40,41,42]. Although promising, the implications of this in terms of
clinical outcomes require confirmation from future clinical trials.
Ventilation and the open lung
The stabilizing properties of surfactant create hysteresis in the breathing,
or ventilated, lung. This is manifested by two limbs (inflation and deflation)
on the static P-V curve. Ideally, one should attempt to inflate the lung so that
cyclic ventilation occurs on the deflation limb of its P-V curve at which lung
volumes are higher at a given pressure [43–46]. The idea behind open-lung
254 J.M. Singh, T.E. Stewart / Respir Care Clin 8 (2002) 247–260
ventilation is to maintain the lung on its deflation limb (Fig. 4). Theoreti-
cally, such a strategy improves lung volumes and increases the number of
aerated alveoli. Consequently, this has a positive effect on oxygenation and
increases respiratory-system compliance, which often allows for ventilation
at lower driving pressures.
Lung recruitment and lung-volume maintenance is particularly importan