CLASSIFICATION OF 2

MECHANICAL VENTILATORS
AND MODES OF VENTILATION
Robert L. Chatburn

CONTROL SYSTEM MODES OF VENTILATION

Models of Patient–Ventilator Interaction Control Variable
Control Variables Breath Sequence
Phase Variables Targeting Schemes
Trigger Variable Mode Classification
Target Variable VENTILATOR ALARM SYSTEMS
Cycle Variable
Baseline Variable THE FUTURE
SUMMARY AND CONCLUSION

A good ventilator classification scheme describes how venti- chapter; these topics have been treated elsewhere.7,8 The
lators work in general terms, but with enough detail so that chapter does, however, explore in detail control schemes
one particular model can be distinguished from others. It and ventilator modes because these directly affect patient
facilitates description by focusing on key attributes in a logi- management.
cal and consistent manner. A clear description allows us to
quickly assess new facts in relation to our previous knowl-
edge. Learning the operation of a new ventilator or describ- CONTROL SYSTEM
ing it to others then becomes much easier. Understanding
how the ventilator operates, we can then anticipate appropri- Models of Patient–Ventilator Interaction
ate ventilator management strategies for particular clinical
situations. The classification system described in this chapter To understand how a machine can be controlled to replace
is based on previously published work.1–7 or supplement the natural function of breathing, we need to
A ventilator is simply a machine, a system of related ele- first understand something about the mechanics of breath-
ments designed to alter, transmit, and direct energy in a pre- ing itself. The study of mechanics deals with forces, displace-
determined manner to perform useful work. We put energy ments, and the rate of change of displacement. In physiology,
into the ventilator in the form of electricity (energy = volts force is measured as pressure (pressure = force/area), dis-
× amps × time) or compressed gas (energy = pressure × placement as volume (volume = area × displacement), and
volume). That energy is transmitted or transformed (by the the relevant rate of change as flow [average flow = Δvolume/
ventilator’s drive mechanism) in a predetermined manner Δtime; instantaneous flow (V ) = dv /dt , the derivative of vol-
(by the control circuit) to augment or replace the patient’s ume with respect to time]. Specifically, we are interested in
muscles in performing the work of breathing. Thus to the pressure necessary to cause a flow of gas to enter the air-
understand mechanical ventilators in general, we must first way and increase the volume of the lungs.
understand their basic functions: (a) power input, (b) power The study of respiratory mechanics is essentially the
transmission or conversion, (c) control scheme, and (d) out- search for simple but useful models of respiratory system
put. This simple format can be expanded to add as much mechanical behavior. Figure 2-1 illustrates the process by
detail as desired (Table 2-1). which the respiratory system is represented first by a graphi-
A discussion of input power sources and power con- cal model, and then by a mathematical model based on the
version and transmission is beyond the scope of this graphical model. Pressure, volume, and flow are measurable

45
46 Part II Physical Basis of Mechanical Ventilation

Resistive pressure is the product of resistance (R = Δpressure/

TABLE 2-1: OUTLINE OF VENTILATOR
Δflow) and flow. Thus, Eq. (1) can be expanded to yield the
CLASSIFICATION SYSTEM
following equation for inspiration:
I. Input IV. Output ˙
Pvent + Pmus = EV + RV
A. Pneumatic A. Pressure waveforms (2)
B. Electri 1. Rectangular
1. AC 2. Exponential
The combined ventilator and muscle pressure causes volume
2. DC (battery) 3. Sinusoidal and flow to be delivered to the patient. (Of course, muscle
II. Power conversion and 4. Oscillating pressure may subtract rather than add to ventilator pres-
transmission B. Volume waveforms sure in the case of patient–ventilator dyssynchrony, in which
A. External compressor 1. Ascending ramp case both volume and flow delivery are reduced.) Pressure,
B. Internal compressor 2. Sinusoidal volume, and flow are functions of time and are called vari-
C. Output control C. Flow waveforms ables. They are all measured relative to their values at end-
valves 1. Rectangular
III. Control scheme 2. Ascending ramp
expiration. Elastance and resistance are assumed to remain
A. Control circuit 3. Descending ramp constant and are called parameters.
1. Mechanical 4. Sinusoidal For passive expiration, both ventilator and muscle pres-
2. Pneumatic V. Alarms sure are absent, so Eq. (2) becomes
3. Fluidic A. Input power alarms
4. Electric 1. Loss of electric power − RV = EV (3)
5. Electronic 2. Loss of pneumatic power
B. Control variables B. Control circuit alarms The negative sign on the left side of the equation indicates
1. Pressure 1. General systems failure flow in the expiratory direction. This equation also shows
2. Volume 2. Incompatible ventilator that passive expiratory flow is generated by the energy stored
3. Time settings
in the elastic compartment (i.e., lungs and chest wall) during
C. Phase variables 3. Warnings (e.g., inverse
1. Trigger inspiratory-to-expiratory inspiration.
2. Target timing ratio) Equation (2) shows that if the patient’s respiratory mus-
3. Cycle C. Output alarms (high/low cles are not functioning, muscle pressure is zero, and the
4. Baseline conditions) ventilator must generate all the pressure for inspiration.
D. Modes of ventilation 1. Pressure On the other hand, a ventilator is not needed for normal
1. Control variable 2. Volume spontaneous breathing (i.e., vent pressure = 0). Between
2. Breath sequence 3. Flow
3. Targeting 4. Time
those two extremes, an infinite number of combinations of
schemes a. Frequency muscle pressure (i.e., patient effort) and ventilator pressure
b. Inspiratory time are possible under the general heading of “partial ventila-
c. Expiratory time tor support.” The equation of motion also gives the basis for
5. Inspired gas defining an assisted breath as one for which ventilator pres-
a. Temperature sure rises above baseline during inspiration or falls below
b. FIO2
baseline during expiration.

Control Variables
variables in the mathematical model that change with time In the equation of motion, the mathematical form of any of
over the course of one inspiration and expiration. The rela- the three variables (i.e., pressure, volume, or flow as func-
tion among them is described by the equation of motion for tions of time) can be predetermined, making it the inde-
the respiratory system.9 The derivation of this equation stems pendent variable and making the other two the dependent
from a force-balance equation that is an expression of New- variables. We now have a theoretical basis for classifying ven-
ton’s third law of motion (for every action, there is an equal tilators as pressure, volume, or flow controllers. Thus, during
and opposite reaction): pressure-controlled ventilation, pressure is the independent
variable and may take the form of, say, a step function (i.e.,
PTR = PE + PR (1)
a rectangular pressure waveform). The shapes of the vol-
where PTR is the transrespiratory pressure (i.e., pressure ume and flow waveforms for a passive respiratory system
at the airway opening minus pressure at the body surface), (Pmus = 0) then depends on the shape of the pressure wave-
PE is the pressure caused by elastic recoil (elastic load), and form as well as the parameters of resistance and compliance.
PR is the pressure caused by flow resistance (resistive load). On the other hand, during volume-controlled ventilation, we
Transrespiratory pressure can have two components, can specify the shape of the volume waveform making flow-
one generated by the ventilator (P vent) and one generated by dependent and pressure-dependent variables. The same rea-
the respiratory muscles (Pmus). Elastic recoil pressure is the soning applies to a flow controller. Notable exceptions are
product of elastance (E = Δpressure/Δvolume) and volume. interpulmonary percussive ventilation, and high-frequency
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 47

Flow

Transairway
pressure

Transrespiratory
pressure
Transthoracic
pressure
Volume

FIGURE 2-1 The respiratory system is often modeled as a single flow resistance (representing the endotracheal tube and the airways) connected
to an elastic chamber (representing the lungs and chest wall). Flow through the airways is generated by transairway pressure (pressure at the airway
opening minus pressure in the lungs). Expansion of the elastic chamber is generated by transthoracic pressure (pressure in the lungs minus pressure
on the body surface). Transrespiratory pressure (pressure at the airway opening minus pressure on the body surface) is the sum of these two pressures
and is the total pressure required to generate inspiration. The “airway-pressure” gauge on a positive-pressure ventilator displays transrespiratory
pressure.

oscillatory ventilation, both of which control only the dura- A particular variable is measured and used to start, sustain,
tion of flow pulses; the resulting airway pressure pulses and end each phase. In this context, pressure, volume, flow,
along with actual inspiratory flows and volumes depend on and time are referred to as phase variables.11 Figure 2-2 shows
the instantaneous values of respiratory system impedance. the criteria for determining phase variables.
Because neither pressure, volume, nor flow in the equation
of motion are predetermined, we would classify this type of
device as a “time controller.” Trigger Variable
It follows from the preceding discussion that any conceiv-
able ventilator can control only one variable at a time: pres- All ventilators measure one or more variables associated with
sure, volume, or flow. Because volume and flow are inverse the equation of motion (i.e., pressure, volume, flow, or time).
functions of one another, we can simplify our discussion Inspiration is started when one of these variables reaches a
and consider only pressure and volume as control variables. preset value. Thus, the variable of interest is considered an
I discuss later in “Modes of Ventilation” exactly how ventila- initiating, or trigger, variable. Time is a trigger variable when
tor control systems work. We will see that it is possible for the ventilator starts a breath according to a set frequency
a ventilator to switch quickly from one control variable to independent of the patient’s spontaneous efforts. Pressure is
another, not only from breath to breath, but even during a the trigger variable when the ventilator senses a drop in base-
single inspiration. line pressure caused by the patient’s inspiratory effort and
begins a breath independent of the set frequency. Flow or
volume are the trigger variables when the ventilator senses
Phase Variables the patient’s inspiratory effort in the form of either flow of
volume into the lungs.
Because breathing is a periodic event, the ventilator must Flow triggering reduces the work the patient must per-
be able to control a number of variables during the respira- form to start inspiration.12 This is so because work is pro-
tory cycle (i.e., the time from the beginning of one breath portional to the volume the patient inspires times the
to the beginning of the next). Mushin et al10 proposed that change in baseline pressure necessary to trigger. Pressure
this time span be divided into four phases: the change from triggering requires some pressure change and hence an
expiration to inspiration, inspiration, the change from inspi- irreducible amount of work to trigger. With flow or volume
ration to expiration, and expiration. This convention is triggering, however, baseline pressure need not change, and
useful for examining how a ventilator starts, sustains, and theoretically, the patient need do no work on the ventilator
stops an inspiration and what it does between inspirations. to trigger.
48 Part II Physical Basis of Mechanical Ventilation

Inspiration is Inspiration is Inspiration is Inspiration is

Pressure triggered Volume triggered Flow triggered Time triggered

Yes Yes Yes

Does inspiration Does inspiration Does inspiration Inspiration starts

start because a preset No start because a preset No start because a preset No because a preset time
pressure is detected? volume is detected? flow is detected? interval has elapsed.
Observation and previous knowledge

Inspiration is Inspiration is Inspiration is

Pressure targeted Volume targeted Flow targeted

Yes Yes Yes

Does peak pressure Does peak volume Does peak flow reach No variables are
reach a preset value No reach a preset value No a preset value before No targeted during
before inspiration before inspiration inspiration ends? inspiration.
ends? ends?

Inspiration is Inspiration is Inspiration is Inspiration is

Pressure cycled Volume cycled Flow cycled Time cycled

Yes Yes Yes

Does expiration start Does expiration start Does expiration start Expiration starts
because a preset No because a preset No because a preset No because a preset
pressure is met? volume is met? flow is met? time is met.
FIGURE 2-2 Criteria for determining the phase variables during a ventilator-assisted breath.

The patient effort required to trigger inspiration is Organization’s use of the term limit as applying to alarm situ-
determined by the ventilator’s sensitivity setting. Some ven- ations only.
tilators indicate sensitivity qualitatively (“min” or “max”). Clinicians often confuse target variables with cycle vari-
Alternatively, a ventilator may specify a trigger threshold ables. To cycle means “to end inspiration.” A cycle variable
quantitatively (e.g., 5 cm H2O below baseline). Once the always ends inspiration. A target variable does not terminate
trigger variable signals the start of inspiration, there is inspiration; it only sets an upper bound for pressure, volume,
always a short delay before flow to the patient starts. This or flow (Fig. 2-3).
delay is called the response time and is secondary to the
signal-processing time and the mechanical inertia of the
drive mechanisms. It is important for the ventilator to have Cycle Variable
a short response time to maintain optimal synchrony with
patient inspiratory effort. The inspiratory phase always ends when some variable
reaches a preset value. The variable that is measured and
used to end inspiration is called the cycle variable. The cycle
Target Variable variable can be pressure, volume, flow, or time. Manual
cycling is also available on some ventilators.
Here target means restricting the magnitude of a variable When a ventilator is set to pressure cycle, it delivers flow
during inspiration. A target variable is one that can reach and until a preset pressure is reached, at which time inspiratory
maintain a preset level before inspiration ends (i.e., it does flow stops and expiratory flow begins. The most common
not end inspiration). Pressure, flow, or volume can be tar- application of pressure cycling on mechanical ventilators is
get variables and actually all can be active for a single breath for alarm settings.
(e.g., using the Pmax feature on a Dräger ventilator). Note that When a ventilator is set to volume cycle, it delivers flow
time cannot be a target variable because specifying an inspi- until a preset volume has passed through the control valve.
ratory time would cause inspiration to end, violating the pre- By definition, as soon as the set volume is met, inspiratory
ceding definition. Astute readers may notice that in the past flow stops and expiratory flow begins. If expiration does not
I have used the term limit where here I have used target. This begin immediately after inspiratory flow stops, then an inspi-
was done to be consistent with the International Standards ratory hold has been set, and the ventilator is, by definition,
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 49

A B C

Ventilator
pressure
Volume
Flow

FIGURE 2-3 This figure illustrates the distinction between the terms target and cycle. A. Inspiration is pressure-targeted and time-cycled. B. Flow is
targeted, but volume is not, and inspiration is volume-cycled. C. Both volume and flow are targeted, and inspiration is time-cycled. (Reproduced, with
permission, from Chatburn.6)

time cycled (see Fig. 2-3). Note that the volume that passes Time cycling means that expiratory flow starts because a
through the ventilator’s output control valve is never exactly preset inspiratory time interval has elapsed.
equal to the volume delivered to the patient because of the
volume compressed in the patient circuit. Some ventilators Baseline Variable
use a sensor at the Y-connector (such as the Dräger Evita
4 with the neonatal circuit) for more accurate tidal volume The baseline variable is the parameter controlled during expi-
measurement. Others measure volume at some point inside ration. Although pressure, volume, or flow could serve as the
the ventilator, and the operator must know whether the ven- baseline variable, pressure control is the most practical and
tilator compensates for compressed gas in its tidal volume is implemented by all modern ventilators. Baseline or expi-
readout. ratory pressure is always measured and set relative to atmo-
When a ventilator is set to flow cycle, it delivers flow spheric pressure. Thus, when we want baseline pressure to
until a preset level is met. Flow then stops, and expiration equal atmospheric pressure, we set it to zero. When we want
begins. The most frequent application of flow cycling is in baseline pressure to exceed atmospheric pressure, we set a
the pressure-support mode. In this mode, the control vari- positive value, called positive end-expiratory pressure (PEEP).
able is pressure, and the ventilator provides the flow neces-
sary to meet the inspiratory pressure target. In doing so, flow
starts out at a relatively high value and decays exponentially MODES OF VENTILATION
(assuming that the patient’s respiratory muscles are inactive
after triggering). Once flow has decreased to a relatively low The general goals of mechanical ventilation are to pro-
value (such as 25% of peak flow, typically preset by the man- mote safety, comfort, and liberation (Table 2-2).1 Specific
ufacturer), inspiration is cycled off. Manufacturers often set objectives under these goals include ensuring adequate gas
the cycle threshold slightly above zero flow to prevent inspi- exchange, avoiding ventilator induced lung injury, optimiz-
ratory times from getting so long that patient synchrony is ing patient-ventilator synchrony, and minimizing the dura-
degraded. On some ventilators, the flow-cycle threshold may tion of ventilation. The preset pattern of patient-ventilator
be adjusted by the operator to improve patient synchrony. interaction designed to achieve these objectives is referred to
Increasing the flow-cycle threshold decreases inspiratory as a mode of ventilation. Specifically, a mode can be classified
time and vice versa. according to the outline in Table 2-3.2
50 Part II Physical Basis of Mechanical Ventilation

TABLE 2-2: GOALS AND OBJECTIVES OF TABLE 2-3: OUTLINE OF MODE

MECHANICAL VENTILATION CLASSIFICATION SYSTEM
1. Promote safety 1. Primary control variable
a. Optimize ventilation–perfusion of the lung a. Pressure
i. Maximize alveolar ventilation b. Volume
ii. Minimize shunt 2. Breath sequence
b. Optimize pressure–volume curve a. Continuous mandatory ventilation (CMV)
i. Minimize tidal volume b. Intermittent mandatory ventilation (IMV)
ii. Maximize compliance c. Continuous spontaneous ventilation (CSV)
2. Promote comfort 3. Primary targeting scheme
a. Optimize patient–ventilator synchrony a. Set-point
i. Maximize trigger–cycle synchrony b. Dual
ii. Minimize auto-PEEP c. Servo
iii. Maximize flow synchrony d. Adaptive
iv. Coordinate mandatory and spontaneous breaths e. Optimal
b. Optimize work demand versus work delivered f. Intelligent
i. Minimize inappropriate shifting of work from
ventilator to patient 4. Secondary targeting scheme
a. Set-point
3. Promote liberation b. Servo
a. Optimize the weaning experience c. Adaptive
i. Minimize adverse events d. Optimal
ii. Minimize duration of ventilation e. Intelligent
Reproduced with permission from Chatburn RL, Mireles-Cabodevila E. Closed
loop control of mechanical ventilation. Respir Care. 2011;56(1):85–98.

flow changes caused by hiccups or cardiogenic oscillations.

Control Variable It allows, however, the superimposition of, for example, a
spontaneous breath on a mandatory breath or vice versa. The
I have already mentioned that pressure, volume, or flow can flows are paired by size, not necessarily by timing. In air-
be controlled during inspiration. When discussing modes way pressure-release ventilation, for example, there is a large
I will refer to inspiration as being pressure-controlled or inspiration (transition from low pressure to high pressure)
volume-controlled. Ignoring flow control is justified because possibly followed by a few small inspirations and expira-
when the ventilator controls volume directly (i.e., using a tions, followed finally by a large expiration (transition from
volume-feedback signal), flow is controlled indirectly, and high pressure to low pressure). These comprise several small
vice versa (i.e., mathematically, volume is the integral of flow, spontaneous breaths superimposed on one large manda-
and flow is the derivative of volume). tory breath. During high-frequency oscillatory ventilation,
There are clinical advantages and disadvantages to vol- in contrast, small mandatory breaths are superimposed on
ume and pressure control. To keep within the scope of this larger spontaneous breaths.
chapter, we can just say that volume control results in a more A spontaneous breath, in the context of mechanical ven-
stable minute ventilation (and hence more stable blood gases) tilation, is a breath for which the patient determines both
than pressure control if lung mechanics are unstable. On the the timing and the size. The start and end of inspiration may
other hand, pressure control allows better synchronization be determined by the patient, independent of any machine
with the patient because inspiratory flow is not constrained settings for inspiratory time and expiratory time. That is,
to a preset value. Although the ventilator must control only the patient both triggers and cycles the breath. On some
one variable at a time during inspiration, it is possible to ventilators, the patient may make short, small spontaneous
begin a breath-in pressure control and (if certain criteria are efforts during a longer, larger mandatory breath, as in the
met) switch to volume control or vice versa (referred to as case of airway pressure-release ventilation. It is important to
dual targeting, described in “Targeting Schemes” below). make a distinction between spontaneous breaths and assisted
breaths. An assisted breath is one for which the ventilator
does some work for the patient, as indicated by an increase
Breath Sequence in airway pressure (i.e., Pvent) above baseline during inspira-
tion or below baseline during expiration. For example, in the
The breath sequence is the pattern of mandatory or sponta- pressure-support mode, each breath is assisted because airway
neous breaths that the mode delivers. A breath is a positive pressures rise to the pressure-support setting above PEEP (i.e.,
airway flow (inspiration) relative to baseline, and it is paired Pvent > 0). Each breath is also spontaneous because the patient
(by size) with a negative airway flow (expiration), both asso- both triggers and cycles the breath. The patient may cycle the
ciated with ventilation of the lungs. This definition excludes breath in the pressure-support mode by actively exhaling, but
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 51

even if the patient is passive at end-inspiration, the patient’s controlled mandatory breaths on infant ventilators. The
resistance and compliance determine the cycle point and thus key distinction between CMV and IMV is that with CMV,
the size of the breath for a given pressure-support setting. In the ventilator attempts to deliver a mandatory breath every
contrast, for a patient on continuous positive airway pressure, time the patient makes an inspiratory effort (unless a man-
each breath is spontaneous but unassisted. Breaths are spon- datory breath is already in progress). This means that
taneous because the patient determines the timing and size of during CMV, if the operator decreases the ventilator rate,
the breaths without any interference by the ventilator. Breaths the level of ventilator support is unaffected as long as the
during continuous positive airway pressure are not assisted patient continues making inspiratory efforts. With IMV,
because airway pressure is controlled by the ventilator to be the rate setting directly affects the number of mandatory
as constant as possible (i.e., Pvent = 0). Understanding the dif- breaths and hence the level of ventilator support. Thus,
ference between assisted and unassisted spontaneous breaths CMV is normally viewed as a method of “full ventilator
is very important clinically. When making measurements of support,” whereas IMV is usually viewed as a method of
tidal volume and respiratory rate for calculation of the rapid- partial ventilator support. Of course, actual “full ventila-
shallow breathing index, for example, the breaths must be tory support” can only be achieved if the patient is making
spontaneous and unassisted. If they are assisted (e.g., with no inspiratory efforts, for example, is paralyzed, but the
pressure support), an error of 25% to 50% may be introduced. term is often used loosely to mean supplying as much sup-
A mandatory breath is any breath that does not meet the port as possible for a given patient condition.
criteria of a spontaneous breath, meaning that the patient
Given the two ways to control inspiration (i.e., pressure
has lost control over the timing and/or size. Thus, a manda-
and volume) and the three breath sequences (i.e., CMV, IMV,
tory breath is one for which the start or end of inspiration
or CSV), there are five possible breathing patterns; volume
(or both) is determined by the ventilator, independent of the
control (VC)-CMV, VC-IMV, pressure control (PC)-CMV,
patient; that is, the machine triggers and/or cycles the breath.
PC-IMV, PC-CSV (see Table 2-2). VC-CSV is not possible
It is possible to superimpose a short mandatory breath on
because volume control implies that inspiration ends after a
top of a longer spontaneous breath, as in the case of high-
preset tidal volume is delivered, hence violating the patient
frequency oscillatory ventilation.
cycling criterion of a spontaneous breath.
Having defined spontaneous and mandatory breaths, there
are three possible breath sequences, designated as follows:
• Continuous spontaneous ventilation (CSV). All breaths are Targeting Schemes
spontaneous.
• Intermittent mandatory ventilation (IMV). Spontane- Targeting schemes are feedback control systems used by
ous breaths are permitted between mandatory breaths. mechanical ventilators to deliver specific ventilatory pat-
When the mandatory breath is triggered by the patient, terns.1 The targeting scheme is a key component of a mode
it is commonly referred to as synchronized IMV. Because classification system. Before we can describe specific target-
the trigger variable can be specified in the description of ing schemes used by ventilators, we must first appreciate the
phase variables, I will use IMV instead of synchronized basic concepts of engineering control theory.
IMV to designate general breath sequences. The term closed-loop control refers to the use of a feed-
• Continuous mandatory ventilation (CMV). Spontaneous back signal to adjust the output of a system. Ventilators use
breaths are not permitted between mandatory breaths, as closed-loop control to maintain consistent pressure and flow
the intent is to provide a mandatory breath for every patient waveforms in the face of changing patient/system condi-
inspiratory effort. CMV originally meant that every breath tions. This is accomplished by using the output as a feedback
was mandatory. The development of the “active exhalation signal that is compared to the operator-set input. The differ-
valve,” however, made it possible for the patient to breathe ence between the two is used to drive the system toward the
spontaneously during a mandatory pressure-controlled desired output. For example, pressure-control modes use
breath on some ventilators. In fact, it was always possible airway pressure as the feedback signal to control gas flow
for the patient to breathe spontaneously during pressure- from the ventilator. Figure 2-4 is a schematic of a general

Disturbances

Error Controller Effector Manipulated Controlled

Input Plant
+ signal (Software) (Hardware) variable variable
– (Output)

Feedback signal
FIGURE 2-4 Generalized control circuit (see text for explanation). The “plant” in a control circuit for mechanical ventilation is the patient.
(Reproduced with permission from Chatburn RL. Mireles-Cabodevila E, Closed loop control of mechanical ventilation. Respir Care. 2011;56(1):
85–98.)
52 Part II Physical Basis of Mechanical Ventilation

control system. The input is a reference value (e.g., operator (relationship between the input and the output of the con-
preset inspiratory pressure) that is compared to the actual troller) as a targeting scheme. The history of these schemes
output value (e.g., instantaneous value of airway pressure). clearly shows an evolutionary trend toward increasing lev-
The difference between those two values is the error signal. els of automation. In fact, we can identify three groups of
The error signal is passed to the controller (e.g., the software targeting schemes based on increasing levels of autonomy:
control algorithm). The controller converts the error signal manual, servo, and automatic. Manual targeting schemes
into a signal that can drive the effector (e.g., the hardware) require the operator to adjust all the target values. Servo
to cause a change in the manipulated variable (e.g., inspira- targeting schemes are unique in that there are no static
tory flow). The relationship between the input and the out- target values; rather, the operator sets the parameters of a
put of the controller is called the transfer function in control mathematical model that drives the ventilator’s output to
theory. Engineers need to understand the transfer function follow a dynamic signal (like power steering on an auto-
in terms of complex mathematical equations. Clinicians, mobile). Automatic targeting schemes enable the ventila-
however, need only understand the general operation of tor to set some or all of the ventilatory targets, using either
the function in terms of how the mode affects the patient’s mathematical models of physiologic processes or artificial-
ventilatory pattern, and we will use that frame of reference intelligence algorithms.
in defining targeting schemes. The “plant” in Figure 2-4 The basic concept of closed-loop control has evolved into
refers to the process under control. In our case, the plant is at least six different ventilator targeting schemes (set-point,
the patient and the delivery circuit connecting the patient dual, servo, adaptive, optimal, and intelligent). These tar-
to the ventilator. The plant is the source of the “noise” that geting schemes are the foundation that makes possible sev-
causes problems with patient–ventilator synchrony. At one eral dozen apparently different modes of ventilation. Once
extreme, a paralyzed patient and an intact delivery circuit we understand how these control types work, many of the
pose little challenge for a modern ventilator to deliver a apparent differences are seen to be similarities. We then
predetermined ventilatory pattern, and thus synchrony is avoid a lot of the confusion surrounding ventilator market-
not an issue. At the opposite extreme is a patient with an ing hype and begin to appreciate the true clinical capabilities
intense, erratic respiratory drive and a delivery circuit with of different ventilators.
leaks (e.g., around an uncuffed endotracheal tube) making
patient–ventilator synchrony virtually impossible. The chal- SET-POINT
lenge for both clinicians and engineers is to develop tech-
nology and procedures for dealing with this wide range of In set-point targeting, the operator sets specific target values
circumstances. and the ventilator attempts to deliver them (Fig. 2-5). The
The plant alters the manipulated variable to generate the simplest examples for volume-control modes are tidal vol-
feedback signal of interest as the control (output) variable. ume and inspiratory flow. For pressure-control modes, the
Continuing with the example above, the manipulated vari- operator may set inspiratory pressure and inspiratory time
able is flow, but the feedback control variable is pressure (i.e., or cycle threshold.
ventilator flow times plant impedance equals airway pres-
sure), as in pressure-control modes. DUAL
Closed-loop control can also refer to the use of feedback As it relates to mechanical ventilation, volume control means
signals to control the overall pattern of ventilation, beyond that inspired volume, as a function of time, is predetermined
a single breath, such as the use of end-tidal carbon dioxide by the operator before the breath begins. In contrast, pres-
tension as a feedback signal to control minute ventilation. sure control means that inspiratory pressure as a function of
The process of “setting” or adjusting a ventilation mode time is predetermined. “Predetermined” in this sense means
can be thought of as presetting various target values, such as that either pressure or volume is constrained to a specific
tidal volume, inspiratory flow, inspiratory pressure, inspira- mathematical form. In the simple case where either pressure
tory time, frequency, PEEP, oxygen concentration, and end- or flow are preset constant values (e.g., set-point targeting, as
tidal carbon dioxide concentration. The term target is used explained above), we can say that they are the independent
for two reasons. First, just like in archery, a target is aimed
at but not necessarily hit, depending on the precision of the
control system. An example is setting a target value for tidal
Disturbances
volume and allowing the ventilator to adjust the inspiratory
pressure over several breaths to finally deliver the desired Set-point
Operator
value. In this case, we could more accurately talk about deliv- Pressure Ventilator Patient
Volume
ering an average target tidal volume over time. Flow
The second reason for using target is because the term Flow or volume
control is overused and we need it to preserve some funda- Pressure
mental conventions regarding modes such as volume control FIGURE 2-5 Set-point targeting. (Reproduced, with permission, from
versus pressure control. From this use of the term target, we Chatburn RL. Computer control of mechanical ventilation. Respir Care.
can logically refer to the control system transfer function 2004;49:507–515.)
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 53

variables in the equation of motion. The equation of motion provides the safety of a guaranteed minimum tidal volume
for the respiratory system is a general mathematical model of with the patient comfort of flow synchrony provided by pres-
patient–ventilator interaction: sure control.
P(t ) = EV (t ) + RV (t ) (4)
SERVO
where P(t) is inspiratory pressure as a function of time (t), E The term servo was coined by Joseph Farcot in 1873 to
is respiratory-system elastance, V(t) is volume as a function describe steam-powered steering systems. Later, hydraulic
of time, R is respiratory-system resistance, and V is flow as a “servos” were used to position antiaircraft guns on warships.
function of time. Thus, for example, if pressure is the inde- Servo control specifically refers to a control system that con-
pendent variable, then both volume and flow are dependent verts a small mechanical motion into one requiring much
variables, indicating pressure control. If volume is the inde- greater power, using a feedback mechanism. As such, it
pendent variable, then pressure is the dependent variable, offers a substantial advantage in terms of creating ventilation
indicating volume control. Because volume is the integral of modes capable of a high degree of synchrony with patient
flow, if V is predetermined, then so is V(t). Therefore, for breathing efforts. That is, ventilator work output can be
simplicity, we include the case of flow being the independent made to match patient work demand with a high degree of
variable as a form of volume control. fidelity. We apply the name servo control to targeting schemes
Only one variable (i.e., pressure or volume) can be inde- in which the ventilator’s output automatically follows a vary-
pendent at any moment, but a ventilator controller can ing input. This includes proportional-assist ventilation (PAV;
switch between the two during a single inspiration. When Fig. 2-6),14 automatic tube compensation (ATC),15 and neu-
this happens, the targeting scheme is called dual set-point rally adjusted ventilatory assist (NAVA),16 in which the air-
control or dual targeting. There are two basic ways that way pressure signal not only follows but amplifies signals
ventilators have implemented dual targeting. One way is to that are surrogates for patient effort (i.e., volume, flow, and
start inspiration in volume control and then switch to pres- diaphragmatic electrical signals). Note that the term servo
sure control if one or more preset thresholds are met (e.g., a control has been loosely used since it was coined to refer to
desired peak airway pressure target). An example of such a any type of general feedback control mechanism, but I am
threshold is the operator-set Pmax in volume control on the using it in a very specific manner, as it applies to ventilator
Dräger Evita XL ventilator. The other form of dual targeting targeting schemes.
is to start inspiration in pressure control and then switch to
volume control (e.g., if a preset tidal volume has not been
met when flow decays to a preset value). This was originally
described as “volume-assured pressure-support ventila- Disturbances
tion,”13 but is currently only available as a mode called “Vol- Set-point
Operator
ume Control Assist Control with Machine Volume” in the Elastic load Ventilator Patient
CareFusion Avea ventilator.
Resistive load
Dual targeting is an attempt to improve the synchrony Pressure, volume, and flow
between patient and ventilator. This can be seen in the equa-
tion of motion if a term representing the patient inspiratory
force (muscle pressure or Pmus) is added: Pmus = Loadnormal + Loaddisease

P(t) = EV(t) + RV(t) − Pmus(t) (5)

.
With set-point targeting in volume control modes, vol- Pvent = K1 × V + K2 × V
ume and flow are preset. Therefore, if the patient makes an
inspiratory effort (i.e., Pmus(t) > 0), then the equation dic-
tates that transrespiratory-system pressure, P(t), must fall. Pmus + Pvent = Loadnormal + Loaddisease
Because work is the result of both pressure and volume deliv-
FIGURE 2-6 Servo targeting is the basis for the proportional-assist
ery (i.e., work = ∫Pdv), if pressure decreases, the work the
mode. In this mode, the operator sets targets for elastic and resis-
ventilator does on the patient decreases and hence we have tive unloading. The ventilator then delivers airway pressure in pro-
asynchrony of work demand on the part of the patient versus portion to the patient’s own inspiratory volume and flow. When the
work output on the part of the ventilator. patient’s muscles have to contend with an abnormal load secondary
With set-point pressure control, transrespiratory pressure to disease, proportional assist allows the operator to set amplifica-
is preset. Consequently, if the patient makes an inspiratory tion factors (K1 and K2) on the feedback volume and flow signals. By
effort, both volume and flow increase. With constant pres- amplifying volume and flow, the ventilator generates a pressure that
supports the abnormal load, freeing the respiratory muscles to sup-
sure and increased volume, work per liter for the breath stays port only the normal load caused by the natural elastance and resis-
constant. Although this gives better work synchrony than tance of the respiratory system. (Reproduced, with permission, from
does volume control, it is not ideal. Nevertheless, merging of Chatburn RL. Computer control of mechanical ventilation. Respir
volume and pressure control using a dual targeting scheme Care. 2004;49:507–515.)
54 Part II Physical Basis of Mechanical Ventilation

Set-point Set-point Model

Operator Minimize
Volume Adjustment
work
Exhaled volume
Pressure
Disturbances Set-point Set-point
Operator
Patient Adjustment

Exhaled volume
weight
Ventilator Patient

Frequency
Pressure

Volume
Flow Disturbances

Pressure
FIGURE 2-7 Adaptive targeting. Notice that the operator has stepped Ventilator Patient
back from direct control of the within-breath parameters of pressure
and flow. Examples of adaptive targeting are pressure-regulated volume Flow
control (PRVC) on the Siemens ventilator and autoflow on the Dräger
Evita 4 ventilator. (Reproduced, with permission, from Chatburn RL. Pressure
Computer control of mechanical ventilation. Respir Care. 2004;49: FIGURE 2-8 Optimal targeting. A static mathematical model is used
507–515.) to optimize some performance parameter, such as work of breath-
ing. The only commercially available form of optimal targeting is the
adaptive-support ventilation (ASV) mode on the Hamilton Galileo ven-
tilator. (Reproduced, with permission, from Chatburn RL. Computer
ADAPTIVE control of mechanical ventilation. Respir Care. 2004;49:507–515.)
An adaptive targeting scheme involves modifying the func-
tion of the controller to cope with the fact that the system
parameters being controlled are time varying. As it applies breathing, mimic natural breathing, stimulate spontaneous
to mechanical ventilation, adaptive targeting schemes allow breathing, and reduce weaning time.20 The operator inputs
the ventilator to set some (or conceivably all) of the targets the patient’s weight. From that, the ventilator estimates the
in response to varying patient conditions. Modern inten- required minute alveolar ventilation, assuming a normal
sive care unit ventilators may use adaptive flow targeting as dead space fraction. Next, an optimum frequency is calcu-
a more accurate way to deliver volume control modes than lated based on work by Otis et al21 that predicts a frequency
set-point targeting. For example, the Covidien PB 840 venti- resulting in the least mechanical work rate:20
lator automatically adjusts inspiratory flow between breaths
MV − f VD ⎛
to compensate for volume compression in the patient circuit −1 + 1 + 4π 2 RC E ⎛
and thus achieving an average target tidal volume equal to ⎝ VD ⎝
the operator-set value.17 Aside from this application of adap- f = (6)
2π 2 RC E
tive targeting, there are four distinct approaches to basic
adaptive targeting, which are represented by the mode names where MV is predicted minute ventilation (L/min) based on
pressure-regulated volume control (inspiratory pressure auto- patient weight and the setting for percent of predicted MV
matically adjusted to achieve an average tidal volume target, to support, VD is predicted dead space (L) based on patient
Fig. 2-7), mandatory rate ventilation (inspiratory pressure weight, RCE is the expiratory time constant calculated as the
automatically adjusted to maintain a target spontaneous slope of the expiratory flow volume curve and f is the com-
breath frequency), adaptive flow/adaptive I-time (inspiratory puted optimal frequency (breaths/min). The target tidal vol-
time and flow automatically adjusted to maintain a constant ume is calculated as MV/f. The ASV controller uses the Otis
inspiratory time-to-expiratory time ratio of 1:2), and manda- equation to set the tidal volume (Fig. 2-8). As with simple
tory minute ventilation (automatic adjustment of mandatory adaptive pressure targeting, the inspiratory pressure within a
breath frequency to maintain a target minute ventilation). breath is controlled to achieve a constant value and between
breaths the inspiratory pressure is adjusted to achieve a tar-
get tidal volume. Unlike simple adaptive pressure targeting,
OPTIMAL
however, the target is not set by the operator; instead, it is
Optimal targeting is an advanced form of adaptive target- estimated by the ventilator in response to changes in respi-
ing.18 Optimal targeting in this context means that the ven- ratory-system mechanics and patient effort. Individual pres-
tilator controller automatically adjusts the targets of the sure-targeted breaths may be mandatory (time triggered and
ventilatory pattern to either minimize or maximize some time cycled) or spontaneous (flow triggered and flow cycle).
overall performance characteristic (Fig. 2-8). Adaptive- ASV adds some expert rules that put safety limits on fre-
support ventilation (ASV) on the Hamilton ventilators is the quency and tidal volume delivery and reduce the risk of auto-
only commercially available mode to date that uses optimal PEEP. In that sense, this mode may be considered an intelligent
targeting. This targeting scheme was first described by Teh- targeting scheme, or more appropriately, a hybrid system (i.e.,
rani in 199120 and was designed to minimize the work rate of using a mathematical model and artificial intelligence).
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 55

INTELLIGENT tube vs. tracheostomy tube), the type of humidifier (heat

and moisture exchanger vs. a heated humidifier), and the
Intelligent targeting systems are another form of adaptive tar-
use of automatic tube compensation. Once the lowest level
geting schemes that use artificial-intelligence techniques.22
of inspiratory pressure is reached, a 1-hour observation
The most convincing proof of the concept was presented by
period is started (i.e., a spontaneous breathing trial) during
East et al,23 who used a rule-based expert system for venti-
which the patient’s breathing frequency, tidal volume, and
lator management in a large, multicenter, prospective, ran-
end-tidal CO2 are monitored. Upon successful completion
domized trial. Although survival and length of stay were not
of this step, a message on the screen suggests that the clini-
different between human and computer management, com-
cian “consider separation” of the patient from the ventilator.
puter control resulted in a significant reduction in multior-
This method for automatic weaning reduces the duration of
gan dysfunction and a lower incidence and severity of lung
mechanical ventilation and intensive care unit length of stay
overdistension injury. The most important finding, however,
in a multicenter randomized controlled trial.24,25 The advan-
was that expert knowledge can be encoded and shared success-
tage of artificial intelligence, however, may be less notice-
fully with institutions that had no input into the model. Note
able in environments where natural intelligence is plentiful.
that the expert system did not control the ventilator directly,
Rose et al recently concluded that “Substantial reductions
but rather made suggestions for the human operator. In the-
in weaning duration previously demonstrated were not
ory, of course, the operator could be eliminated.
confirmed when the SmartCare/PS system was compared
There is only one ventilator mode commercially avail-
to weaning managed by experienced critical care specialty
able to date in the United States with a targeting scheme
nurses, using a 1:1 nurse-to-patient ratio. The effect of
that relies entirely on a rule-based expert system (Fig. 2-9).
SmartCare/PS may be influenced by the local clinical orga-
That mode is SmartCare/PS on the Dräger Evita XL venti-
nizational context.”26
lator. This mode is a specialized form of pressure support
The ultimate in ventilator targeting system to date is the
that is designed for true (ventilator led) automatic weaning
artificial neural network (Fig. 2-10).27 Again, this experimen-
of patients. The SmartCare/PS controller uses predefined
tal system does not control the ventilator directly but acts as
acceptable ranges for spontaneous breathing frequency, tidal
a decision-support system. What is most interesting is that
volume, and end-tidal carbon dioxide tension to automati-
the neural network is capable of learning, which offers signifi-
cally adjust the inspiratory pressure to maintain the patient
cant advantages over static mathematical models and even
in a “respiratory zone of comfort.”23
expert rule-based systems.
The SmartCare/PS system divides the control process
Neural nets are essentially data-modeling tools used to
into three steps. The first step is to stabilize the patient
capture and represent complex input–output relationships.
within the “zone of respiratory comfort” defined as com-
A neural net learns by experience the same way a human
binations of tidal volume, respiratory frequency, and end
brain does, by storing knowledge in the strengths of inter-
tidal CO2 values defined as acceptable by the artificial-
node connections. As data-modeling tools, they have been
intelligence program. There are different combinations
used in many business and medical applications for both
depending on whether the patient has chronic obstructive
diagnosis and forecasting.28 A neural network, like an ani-
pulmonary disease or a neuromuscular disorder. The sec-
mal brain, is made up of individual neurons. Signals (action
ond step is to progressively decrease the inspiratory pressure
potentials) appear at the unit’s inputs (synapses). The effect
while making sure the patient remains in the “zone.” The
of each signal may be approximated by multiplying the sig-
third step tests readiness for extubation by maintaining the
nal by some number or weight to indicate the strength of the
patient at the lowest level of inspiratory pressure. The lowest
signal. The weighted signals then are summed to produce
level depends on the type of artificial airway (endotracheal

Ventilator Disturbances
Patient
Weight Controller Effector
Flow Patient
Diagnosis Expert Rules IP (Hardware)

Inspired flow
Pressure

Volume Integrator Expired flow

Frequency

End tidal CO2

FIGURE 2-9 An intelligent targeting system for automatically adjusting pressure support levels (e.g., SmartCare/PS). IP, inspiratory pressure.
(Reproduced, with permission, from Chatburn RL, Mireles-Cabodevila E. Closed loop control of mechanical ventilation. Respir Care. 2011;56(1):85–98.
56 Part II Physical Basis of Mechanical Ventilation

Single neuron

Threshold
Inputs Weights Summation function Output

1
X Σ
0

Neural network

Input First Second Output

layer hidden hidden layer
layer layer
FIGURE 2-10 Neural network structure. A single neuron accepts inputs of any value and weights them to indicate the strength of the synapse. The
weighted signals are summed to produce an overall unit activation. If this activation exceeds a certain threshold, the unit produces an output response.
A network is made up of layers of individual neurons. (Reproduced, with permission, from Chatburn RL. Computer control of mechanical ventilation.
Respir Care. 2004;49:507–515.)

an overall unit activation. If this activation exceeds a cer- system learning from experience so that the control actually
tain threshold, the unit produces an output response. Large spans between patients instead of just between breaths.
numbers of neurons can be linked together in layers (see
Fig. 2-10). The nodes in the diagram represent the summa-
tion and transfer processes. Note that each node contains Mode Classification
information from all neurons. As the network learns, the
weights change, and thus the values at the nodes change, When Mushin et al wrote the classic book on automatic
affecting the final output. ventilation of the lungs,10 the emphasis was on classifying
In summary, ventilator control schemes display a definite ventilators and there were very few modes on each device.
hierarchy of evolutionary complexity. At the most basic level, These devices have undergone a tremendous technological
control is focused on what happens within a breath. We can evolution during the intervening years. As a result, there are
call this manual control, and there is a very direct need for now more than 170 names of modes on ventilators in the
operator input of static set-points. The next level up is what United States alone, with as many as two dozen available
we can call automatic control. Here, set-points are dynamic on a single device. The proliferation of names makes edu-
in that they may be adjusted automatically over time by the cation of end users very difficult, potentially compromis-
ventilator according to some model of desired performance. ing the quality of patient care. In addition, although there
The operator is somewhat removed in that inputs are entered may be more than 170 mode names, these are not uniquely
at the level of the model and take effect over several breaths different modes. Consequently, the emphasis today in
instead of at the level of individual breath control. Finally, describing ventilators must be on classifying modes, shift-
the highest level so far is what might be considered intelligent ing awareness from names to tags. Much has been written
control. Here, the operator can be eliminated altogether. Not on the subject,2,5, 29–31 and this section gives a brief overview
only dynamic set-points but also dynamic models of desired of the development and application of a ventilator mode
performance are permitted. There is the possibility of the taxonomy.
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 57

You can easily appreciate the motivation for classifying mode can deliver. CSV implies all spontaneous breaths;
modes, just as we do animals or plants (or cars or drugs) IMV allows spontaneous breaths to occur between
because of their large number and variety. The logical basis mandatory breaths and CMV does not.
for a mode taxonomy, however, is not apparent without some 8. There are only five basic ventilatory patterns: VC-CMV,
consideration. This basis has become a teaching system VC-IMV, PC-CMV, PC-IMV, and PC-CSV. All modes
I have developed and tested and is founded on ten simple can be categorizes by these five patterns. This provides
constructs (or aphorisms), each building on the previous one enough practical detail about a mode for most clinical
to yield a practical taxonomy. These aphorisms summarize purposes.
many of the ideas discussed previously in this chapter, and 9. Within each ventilatory pattern there are several variations
there is even some evidence that they are recognized inter- that can be distinguished by their targeting scheme(s).
nationally by clinicians.32 In simplified form, the aphorisms When comparing modes or evaluating the capability of
are as follows: a ventilator, more detail is required than just the ventila-
tory pattern. Modes with the same pattern can be dis-
1. A breath is one cycle of positive flow (inspiration) and neg-
tinguished by describing the targeting schemes they use.
ative flow (expiration). The purpose of a ventilator is to
There are at present only six basic targeting schemes: set-
assist breathing. Therefore, the logical start of a taxonomy
point, dual, servo, adaptive, optimal, and intelligent.
is to define a breath. Breaths are defined such that dur-
10. A mode of ventilation is classified according to its con-
ing mechanical ventilation, small artificial breaths may be
trol variable, breath sequence, and targeting scheme(s).
superimposed on large natural breaths or vice versa.
A practical taxonomy of ventilatory modes is based on
2. A breath is assisted if pressure rises above baseline during
just four levels of detail: the control variable (pressure or
inspiration or falls during expiration. A ventilator assists
volume), the breath sequence (CMV, IMV, or CSV), the
breathing by doing some portion of the work of breath-
targeting scheme used for primary breaths (CMV and
ing. This occurs by delivering volume under pressure.
CSV), and, if applicable, secondary breaths (IMV).
3. A ventilator assists breathing using either pressure con-
trol (PC) or volume control (VC). The equation of In teaching these constructs to respiratory therapists and
motion is the fundamental model for understanding physicians, most educators would agree that knowing a con-
patient–ventilator interaction and hence modes of ven- cept and applying it are two different skills. As with any tax-
tilation. The equation is an expression of the idea that onomy, learning the definitions and mastering the heuristic
only one variable can be predetermined at a time; pres- thinking required to actually categorize specific cases requires
sure or volume (flow control is ignored for simplicity further guidance and some practice. Say, for example, your
and for historical reasons, and because controlling flow task is to compare the capabilities of two major intensive care
directly will indirectly control volume and vice versa). unit ventilator models for a large capital purchase. Memoriz-
4. Breaths are classified according to the criteria that trig- ing the ten aphorisms may not translate into the ability to
ger (start) and cycle (stop) inspiration. A ventilator must classify the modes offered on these two ventilators as a basis
know when to start and stop flow delivery for a given for comparison. To facilitate that skill, I created the three tools
breath. Because starting and stopping inspiratory flow shown in Figures 2-11 and 2-12 and in Table 2-4. Using these
are critical events in synchronizing patient–ventilator tools you can create a simple spreadsheet that defines and
interaction, and because they involve uniquely different compares the modes on any number of ventilators. Table 2-5
operator-influenced factors, they are distinguished by is an example of such a table for the Covidien PB 840 ventila-
giving them different names. tor and the Dräger Evita XL ventilator. When implemented as
5. Trigger and cycle criteria can be either patient or machine a spreadsheet with built-in data-sorting functions, the table
initiated. A major design consideration in creating modes becomes a database with several major uses:
is the ability to synchronize breath delivery with patient
1. A “Rosetta Stone” that can be used to translate from mode
demand and at the same time to guarantee breath delivery
name to mode classification and vice versa. In this way
if the patient is apneic. Therefore, understanding patient–
modes can be identified that are functionally identical but
ventilator interaction means understanding the difference
have different proprietary names.
between machine and patient trigger and cycle events.
2. A tool for engineers to describe performance character-
6. Breaths are classified as spontaneous or mandatory based
istics of individual named modes. Information like this
on both the trigger and cycle criteria. A spontaneous breath
should be available to users in the ventilator’s manual.
arises without apparent external cause. Thus, it is patient
3. A system for clinicians to compare and contrast the capa-
triggered and patient cycled. Any machine involvement
bilities of various modes and ventilators.
in triggering or cycling leads to a mandatory breath. Note
4. A paradigm for educators to use in teaching the basic
that the definition of a spontaneous breath is independent
principles of mechanical ventilation.
of the definition of an assisted or unassisted breath.
7. Ventilators deliver only three basic breath sequences: One can imagine the utility of an expanded database con-
CMV, IMV, and CSV. The two breath classifications taining the classification of all modes on all commercially
logically lead to three possible breath sequences that a available ventilators.
58 Part II Physical Basis of Mechanical Ventilation

Review list of
operator-initiated
Evaluate ventilator
settings and
specifications
ventilator-initiated
settings

Identify what
happens
during a single
breath

VT = Tidal volume
TI = Inspiratory time
Intrabreath
VT is preset Control variable
*Examples:
directly or by flow Yes is
Volume-assist control
and TI* volume
Volume SIMV

Record control
No
variable

Paw = Airway pressure

*Examples: Control variable

CPAP Paw is preset* Yes is
Pressure support pressure
Volume assured pressure support

No

Paw is
proportional to
*Examples: Yes
inspiratory
Automatic tube compensation effort*
Proportional-assist ventilation
Neurally adjusted ventilatory assist

No

Control variable
is
time*

*Example:
Interpulmonary percussive ventilation
FIGURE 2-11 Algorithm for determining the control variable when classifying a mode. SIMV, synchronized intermittent mandatory ventilation.
(Copyright 2011 by Mandu Press Ltd. and reproduced with permission.)

VENTILATOR ALARM SYSTEMS MacIntyre33 has suggested that alarms also be catego-
rized by the events that they are designed to detect. Level
As with other components of ventilation systems, ventilator 1 events include life-threatening situations, such as loss
alarms have increased in number and complexity. Fortu- of input power or ventilator malfunction (e.g., excessive
nately, the classification system I have been describing can or no flow of gas to the patient). The alarms in this cat-
be expanded to include alarms as well (see Table 2-1). egory should be mandatory (i.e., not subject to operator
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 59

Patient trigger variables Trigger = Start Inspiration

- Airway pressure change Cycle = Stop Inspiration
- Inspiratory or expiratory flow change
Review list of - Bioelectrical signal CMV = Continuous Mandatory Ventilation
operator-initiated - Other signal of patient effort IMV = Intermittent Mandatory Ventilation
Evaluate ventilator Patient cycle variables CSV = Continuous Spontaneous Ventilation
settings and
specifications - Airway pressure change APRV = Airway Pressure Release Ventilation
ventilator-initiated
settings - Inspiratory flow change SIMV = Synchronized Intermittent Mandatory
- Bioelectrical signal Ventilation
- Other signal of patient effort HFV = High-Frequency Ventilation
Machine trigger variables Pmus = Ventilatory Muscle Pressure
Identify what - Time (Preset frequency) R = Resistance
happens - Minute ventilation C = Compliance
during a single - Other machine signal independent of patient mechanics (Pmus, R, C)
breath Machine cycle variables
- Time (Preset inspiratory time)
- Volume
- Other machine signal independent of patient mechanics (Pmus, R, C)

Patient Spontaneous Breath

can trigger No breath is not sequence is
inspiration possible CMV

Yes

Patient
can cycle No No
inspiration*
*Normal operation
not alarm condition
Spontaneous
Yes breaths between Yes
mandatory*
*Example
Spontaneous - SIMV
breath is possible
No

Machine trigger Mandatory Unrestricted

Yes spontaneous
possible breath is possible
breathing*
*Examples
- APRV
No - HFV
Yes

Mandatory Breath
Machine cycle
breath is not No Yes sequence is
possible*
possible IMV
*Normal operation
not safety backup feature
Breath
sequence is
CSV

Record breath
sequence

FIGURE 2-12 Algorithm for determining the breath sequence when classifying a mode. (Copyright 2011 by Mandu Press Ltd. and reproduced with
permission.)
 60
TABLE 2-4: EXPLANATION OF HOW TARGETING SCHEMES TRANSFORM OPERATOR INPUTS INTO VENTILATOR OUTPUTS

Predetermined Inputs Ventilator Output

Part II
Control Target
# Variable Scheme Explanation Example Mode Name WB Target Cycle BB Target + Impedance − Impedance

1 P Set-point Peak airway pressure is PC SIMV P T P

Physical Basis of Mechanical Ventilation

independent of impedance
F
2 P Set-point Peak airway pressure is Pressure support P F
independent of impedance

3 P Set-point Peak airway pressure is Automatic resuscitator F P

independent of impedance

4 V Set-point Tidal volume is independent VC A/C F T

of impedance

5 P Dual P-F Same as #1 if secondary target VAPS P ,F V

not activated

6 V Dual F-P Same as #4 if secondary CMV + Pressure F ,P V

target not activated Limited
7 P Servo Pressure is automatically Percent F
proportional to inspiratory effort Support
Effort is represented by patient:
flow ATC
volume and flow PAV+
8 P Servo Pressure is authomatically NAVA cm H2O NA
proportional to inspiratory effort μv
represented by diaphgram EMG Edi Edi

9 P Adaptive Same as #1 within a breath plus PRVC NA T Volume

volume target between breaths

10 P Optimal Same as #9 plus algorithm to ASV NA F %MV

minimize inspiratory work rate Frequency
Volume
11 P Intelligent Same as #9 plus volume, PCO2 and Smart Care/PS NA NA Frequency
frequency targets using artificial Volume
intelligence algorithms PETCO2

P, pressure; V, volume; F, flow; T, time; R, resistance; E, elastance; MV, minute volume; Edi, electrical activity of diaphragm; low impedance, low resistance and/or elastance;
WB Target, within-breath preset parameters of the pressure, volume, or flow waveform; BB Target, between breath targets modify high impedance, high resistance and/or elastance;
WB targets or overal ventiltory pattern; Cycle, end of inspiration; NA, not available as operator preset, ventilator
determines value if applicable.
Source: Copyright 2011 by Mandu Press Ltd, and reproduced with permission.
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 61

TABLE 2-5: SPREADSHEET EXAMPLE OF HOW MODES ON TWO COMMON ICU

VENTILATORS WOULD BE CLASSIFIED
The spreadsheet could be sorted any number of ways (e.g., using AutoFilter drop-down dialogs) to compare the ventilators on various capabilities
(e.g., all modes with adaptive pressure targeting). The spreadsheet also functions as a mode translator, giving the different proprietary names for
identical modes.
Order Family Genus Species
Primary Secondary
Breath Breath
Primary
Manufacturer’s Control Breath Target Target
Manufacturer Model Mode Name Variable Sequence Scheme Scheme
Covidien 840 Volume Control Plus Assist Control Pressure CMV adaptive N/A
Covidien 840 Volume Support Pressure CSV adaptive N/A
Covidien 840 Volume Control Plus Pressure IMV adaptive set-point
Synchronized Intermittent
Mandatory Ventilation
Covidien 840 Volume Ventilation Plus Pressure IMV adaptive adaptive
Synchronized Intermittent
Mandatory Ventilation
Covidien 840 Tube Compensation Pressure CSV servo N/A
Covidien 840 Proportional Assist Plus Pressure CSV servo N/A
Covidien 840 Pressure Control Assist Control Pressure CMV set-point N/A
Covidien 840 Pressure Support Pressure CSV set-point N/A
Covidien 840 Spontaneous Pressure CSV set-point N/A
Covidien 840 Pressure Control Synchronized Pressure IMV set-point set-point
Intermittent Mandatory
Ventilation
Covidien 840 BiLevel Pressure IMV set-point set-point
Covidien 840 Volume Control/Assist Control Volume CMV set-point N/A
Covidien 840 Volume Control Synchronized Volume IMV set-point set-point
Intermittent Mandatory
Ventilation
Dräger Evita XL Mandatory Minute Volume with Pressure IMV adaptive set-point
AutoFlow
Dräger Evita XL Continuous Mandatory Ventilation Pressure CMV adaptive N/A
with AutoFlow
Dräger Evita XL Synchronized Intermittent Pressure IMV adaptive set-point
Mandatory Ventilation with
AutoFlow
Dräger Evita XL SmartCare Pressure CSV intelligent N/A
Dräger Evita XL Automatic Tube Compensation Pressure CSV servo N/A
Dräger Evita XL Pressure Controlled Ventilation Pressure CMV set-point set-point
Plus Assisted
Dräger Evita XL Pressure Controlled Ventilation Pressure IMV set-point set-point
Plus Pressure Support
Dräger Evita XL Airway Pressure Release Ventilation Pressure IMV set-point set-point
Dräger Evita XL Continuous Positive Airway Pressure CSV set-point N/A
Pressure/Pressure Support
Dräger Evita XL Mandatory Minute Volume Volume IMV adaptive set-point
Dräger Evita XL Continuous Mandatory Ventilation Volume CMV dual N/A
with Pressure Limited Ventilation
Dräger Evita XL Synchronized Intermittent Volume IMV dual set-point
Mandatory Ventilation with
Pressure Limited Ventilation
Dräger Evita XL Mandatory Minute Volume with Volume IMV dual/ set-point
Pressure Limited Ventilation adaptive
Dräger Evita XL Continuous Mandatory Ventilation Volume CMV set-point N/A
Dräger Evita XL Synchronized Intermittent Volume IMV set-point set-point
Mandatory Ventilation

CMV, continuous mandatory ventilation; CSV, continuous spontaneous ventilation; IMV, intermittent mandatory ventilation.
Source: Copyright 2011 by Mandu Press Ltd. and reproduced with permission.
62 Part II Physical Basis of Mechanical Ventilation

choice), redundant (i.e., multiple sensors and circuits), difficult. It is like buying a Ferrari and putting wooden
and noncanceling (i.e., alarm continues to be activated, wheels on it. In the future, water vapor should be treated
even if the event is corrected, and must be reset manu- like any other desirable inhaled gas constituent (e.g., air,
ally). Level 2 events can lead to life-threatening situations oxygen, helium, or nitric oxide) and metered from within
if not corrected in a timely fashion. These events include the ventilator. The inspiratory part of the patient circuit
such things as blender failure, high or low airway pressure, should be a sterile, insulated, permanent part of the ven-
autotriggering, and partial patient circuit occlusion. They tilator right up to the patient connection, which can be a
also may include suspicious ventilator settings such as an disposable tip for cleaning purposes. The gas should be
inspiratory-to-expiratory timing (I:E) ratio greater than delivered under high pressure as a jet to provide not only
1:1. Alarms for level 2 events may not be redundant and conventional pressure, volume, and flow waveforms but
may be self-canceling (i.e., alarm inactivated if event ceases also high-frequency ventilation. The jet also can be used
to occur). Level 3 events are those that affect the patient– to provide a counterflow PEEP effect, eliminating any
ventilator interface and may influence the level of support need for an exhalation–valve system. The disposable tip
provided. Examples of such events are changes in patient could be designed to house disposable sensors and would
compliance and resistance, changes in patient respiratory be the only part of the circuit to be exposed to the patient’s
drive, and auto-PEEP. Alarm function at this level is similar exhaled gas. If ventilator manufacturers saw themselves as
to that of level 2 alarms. Level 4 events reflect the patient providers of the entire system, instead of letting third par-
condition alone rather than ventilator function. As such, ties deal in plastic connecting tubing, I think we would see
these events usually are detected by stand-alone monitors, a huge evolutionary step in ventilator performance, better
such as oximeters, cardiac monitors, and blood-gas ana- patient outcomes, and potential savings in labor costs for
lyzers. Some ventilators, however, are able to incorporate providers.
the readings of a capnograph in their displays and alarm Third, the most exciting area for development probably
systems. is in the intelligence that will be built into future ventila-
tor control circuits. The real challenge in closed-loop con-
trol of ventilation is defining, measuring, and interpreting
the appropriate feedback signals. If we stop to consider
THE FUTURE all the variables a human operator assesses, the problem
looks insurmountable. Not only does a human consider a
Almost 20 years ago, Warren Sanborn predicted that ventilators wide range of individual physiologic variables, but there
today would “… report the patient’s metabolic state; manage are the more abstract evaluations of such things as meta-
oxygen delivery; calculate cardiac output, synchronize breath bolic, cardiovascular, and psychological states. Add to this
delivery with cardiac cycle to maximize cardiac output…and the various environmental factors that may affect opera-
perform all these functions automatically or at least presenting tor judgment, and we get a truly complex control problem
consensus-based advisory messages to the practitioner….”17 (Fig. 2-13).
Some of these ideas were never developed commercially. I would like to speculate now about a response to this
Some were tried and abandoned. Some, have evolved beyond challenge. The ideal control strategy would have to start out
Warren’s broad vision. with basic tactical control of the individual breath. Next, we
There are three basic ways to improve ventilators in the add longer-term strategic control that adapts to changing
future. First, just like computer games, ventilators need to load characteristics. Mathematical models could provide the
improve the operator interface constantly. Yet very little basic parameters of the mode, whereas expert rules would
research has been done to call attention to problems with place limits to ensure lung protection.
current displays.34,35 We have come a long way from using a Next, we sample various physiologic parameters and use
crank to adjust the stroke of a ventilator’s piston to set tidal fuzzy logic to establish the patient’s immediate condition.
volume. The operator interface must provide for three basic This information is passed on to a neural network, which
functions: allow input of control and alarm parameters, would then select the best response to the patient’s condition.
monitor the ventilator’s status, and monitor the ventilator– The neural network ideally would have access to a huge
patient interaction status. We have a long way to go before database comprised of both human expert rules and actual
the user interface provides an ideal experience with these patient responses to various ventilator strategies. This
functions. arrangement would allow the ventilator not only to learn
Second, the weak link in the patient–ventilator system from its interaction with the current patient but also to con-
is the patient circuit. We buy a $35,000 ventilator with tribute to the database.
state-of-the-art computer control, and then we connect it Finally, the database and this ventilator could be net-
to the patient (priceless) with a $1.98 piece of plastic tub- worked with other intelligent ventilators to multiply the
ing that is subject to filling with condensate from a heated learning capacity exponentially (Fig. 2-14). Whatever
humidifier whose design has not changed appreciably the future brings, it seems clear that ventilators will have
in 20 years. The resistance and compliance of the deliv- more intelligence built in to increase patient safety and
ery circuit make flow control and volume delivery more decrease the time required to provide care.
 Chapter 2 Classification of Mechanical Ventilators and Modes of Ventilation 63

Set-point
Adjustment

Pressure
Volume
Flow
Pressure Resp rate
Environment
(PIP and PEEP) Heart rate
Volume PeCO2
Frequency PaO2
Time FiO2 FiO2
Cost SpO2
Triage priority P0.1
Experience Disturbances

Alarms
Operator Ventilator Patient

Flow
Pressure

Bronchospasm Metabolic state

Underlying disease Acid–base state
Strength/Endurance Cardiovascular state
Neural control Psychological state
Auto-PEEP Drugs
FIGURE 2-13 The challenge of total computer control of mechanical ventilation. Solid arrows depict signals that have been used at least experimen-
tally. Dotted arrows represent potential feedback signals. (Reproduced, with permission, from Chatburn RL. Computer control of mechanical ventila-
tion. Respir Care. 2004;49:507–515.)

Human
experts
Optimization
models
Strategic
Competitive control
Registry neural
Database network Intelligent Expert
Prior Determine control rules
experience best
rules Disturbances

Ventilator Patient

Networked ventilators Flow

Tactical control
Pressure

Fuzzy
logic
Determine
patient
condition

FIGURE 2-14 A potential approach to the challenge of fully automated control of mechanical ventilation. (Reproduced, with permission, from
Chatburn RL. Computer control of mechanical ventilation. Respir Care. 2004;49:507–515.)
64 Part II Physical Basis of Mechanical Ventilation

SUMMARY AND CONCLUSION 11. Desautels DA. Ventilator performance. In: Kirby RR, Smith RA,
Desautels DA, eds. Mechanical Ventilation. New York, NY: Churchill
Livingstone; 1985:120.
Mechanical ventilators have become so complex that a 12. Sassoon CSH, Girion AE, Ely EA, Light RW. Inspiratory work of
system of classification is necessary to communicate intel- breathing on flow-by and demand flow continuous positive airway
pressure. Crit Care Med. 1989;17:1108–1114.
ligently about them. The theoretical basis for this classifica- 13. Amato MB, Barbas CS, Bonassa J, et al. Volume-assured pres-
tion system is a mathematical model of patient–ventilator sure support ventilation (VAPS): a new approach for reducing
interaction known as the equation of motion for the respira- muscle workload during acute respiratory failure. Chest. 1992;102:
tory system. From this model we deduce that as far as an 1225–1234.
14. Younes M. Proportional assist ventilation, a new approach to ventila-
individual inspiration is concerned, any conceivable venti- tor support: I. Theory. Am Rev Respir Dis. 1992;145:114–120.
lator can be classified as either a pressure, volume, or flow 15. Guttmann J, Eberhard L, Fabry B, et al. Continuous calculation of
controller (and in rare cases, simply an inspiratory-expira- intratracheal pressure in tracheally intubated patients. Anesthesiology.
1993;79(3):503–513.
tory time controller). An individual breath is shaped by the 16. Sinderby C, Beck J, Spahija J, et al. Inspiratory muscle unloading by
phase variables that determine how the breath is triggered neurally adjusted ventilatory assist during maximal inspiratory efforts
(started), targeted (sustained), and cycled (stopped). in healthy subjects. Chest. 2007;131(3):711–717.
A mode of ventilation can be characterized using a four- 17. Sanborn WG. Microprocessor-based mechanical ventilation. Respir
Care. 1993;38(1):72–109.
level taxonomy: (a) control variable, that is, pressure or vol- 18. Stengel RF. Optimal Control and Estimation. Mineola, NY: Dover Pub-
ume according to the equation of motion; (b) the breath lications, 1994.
sequence, that is, CMV, IMV, or CSV; (c) targeting scheme 19. Tehrani FT. Automatic control of an artificial respirator. Conf Proc
for primary breaths; and (d) targeting scheme for second- IEEE Eng Med Biol Soc. 1991;13:1738–1739.
20. Tehrani FT. Automatic control of mechanical ventilation, Part 2:
ary breaths. The trend in ventilator targeting schemes The existing techniques and future trends. J Clin Monit Comput.
has been from basic manual control (within-breath con- 2008;22(6):417–424.
trol requiring operator input of static set-points), to more 21. Otis AB, Fenn WO, Rahn H. Mechanics of breathing in man. J Appl
Physiol. 1950;2:592–607.
advanced automatic control (between-breath control of set- 22. Intelligent control. http://en.wikipedia.org/wiki/Intelligent_control.
points that are adjusted automatically by the ventilator with Last modified October 10, 2011. Last accessed April 30, 2010.
minimal operator input), to the highest level of intelligent 23. East TD, Heermann LK, Bradshaw RL, et al. Efficacy of computerized
control (in which the operator theoretically may be elimi- decision support for mechanical ventilation: results of a prospective
multicenter randomized trial. Proc AMIA Symp. 1999;251–255.
nated altogether in favor of artificial-intelligence systems 24. Lellouche F, Mancebo J, Jolliet P, et al. A multicenter randomized trial
capable of learning). of computer-driven protocolized weaning from mechanical ventila-
tion. Am J Respir Crit Care Med. 2006;174(8):894–900.
25. Rose L, Presneill JJ, Cade JF. Update in computer-driven weaning from
mechanical ventilation. Anaesth Intensive Care. 2007;35:213–221.
26. Rose L, Presneill JL, Johnston L, Cade JF. A randomised, controlled
REFERENCES trial of conventional versus automated weaning from mechanical
ventilation using SmartCare/PS. Intensive Care Med. 2008;34(10):
1. Chatburn RL, Mireles-Cabodevila E. Closed-loop control of mechani- 1788–1795.
cal ventilation: description and classification of targeting schemes. 27. Snowden S, Brownlee KG, Smye SW, Dear PR. An advisory system for
Respir Care. 2011;56(1):85–102. artificial ventilation of the newborn utilizing a neural network. Med
2. Chatburn RL. Understanding mechanical ventilators. Expert Rev Inform (Lond). 1993;18:367–376.
Respir Med. 2010;4(6):809–819. 28. Gottschalk A, Hyzer MC, Greet RT. A comparison of human and
3. Chatburn RL, Volsko TA. Mechanical ventilators. In: Wilkins RL, machine-based predictions of successful weaning from mechanical
Stoller JK, Scanlan CL, eds. Egan’s Fundamentals of Respiratory Care. ventilation. Med Decis Making. 2000;20:243–244.
8th ed. St. Louis, MO: Mosby; 2003:929–962. 29. Chatburn RL. Classification of ventilator modes: update and proposal
4. Chatburn RL. Mechanical ventilators: classification and principles of for implementation. Respir Care. 2007;52(3):301–323.
operation. In: Hess DR, MacIntyre NR, Mishoe SC, et al, eds. Respi- 30. Chatburn RL, Volsko TA. Mechanical ventilators. In: Stoller JK, Kac-
ratory Care: Principles and Practice. Philadelphia, PA: Saunders; marek RM, eds. Egan’s Fundamentals of Respiratory Care. 10th ed. St.
2002:757–809. Louis, MO: Mosby Elsevier; 2011 (in press).
5. Chatburn RL, Primiano FP Jr. A new system for understanding modes 31. Chatburn RL, Volsko TA. Mechanical ventilators: classification and
of mechanical ventilation. Respir Care. 2001;46:604–621. principles of operation. In: Hess DR, MacIntyre NR, Mishoe SC, et
6. Chatburn RL. Fundamentals of Mechanical Ventilation. Cleveland al, eds. Respiratory Care: Principles and Practice. 2nd ed. Philadelphia,
Heights, OH: Mandu Press; 2003. PA: Saunders; 2011 (in press).
7. Branson RD, Hess DR, Chatburn RL. Respiratory Care Equipment. 2nd 32. Chatburn RL. Determining the basis for a taxonomy of mechanical
ed. Philadelphia, PA: Lippincott Williams & Wilkins; 1999. ventilation. Respir Care. 2009;54(11):1555.
8. Cairo JM, Pilbeam SP. Mosby’s Respiratory Care Equipment. 7th ed. St. 33. MacIntyre NR. Ventilator monitors, displays, and alarms. In:
Louis, MO: Mosby; 2004. MacIntyre NR, Branson RD, eds. Mechanical Ventilation. Philadelphia,
9. Rodarte JR, Rehder K. Dynamics of respiration. In: Macklem PT, PA: Saunders; 2001:131–144.
Mead J, eds. Handbook of Physiology. Section 3: The Respiratory System. 34. Wachter SB, Johnson K, Albert R, et al. The evaluation of a pulmo-
Volume. III: Mechanics of Breathing. Part 1. Bethesda, MD: American nary display to detect adverse respiratory events using high resolution
Physiological Society; 1986;131–144. human simulator. J Am Med Inform Assoc. 2006;13(6):635–642.
10. Mushin WW, Rendell-Baker L, Thompson PW, Mapelson WW. Auto- 35. Uzawa Y, Yamada Y, Suzukawa M. Evaluation of the user interface
matic Ventilation of the Lungs. 3rd ed. Oxford, England: Blackwell Sci- simplicity in the modern generation of mechanical ventilators. Respir
entific; 1980:62–131. Care. 2008;53(3):329–337.