Breathing Frequency and Pattern Are
Poor Predictors of Work of Breathing in
Patients Receiving Pressure Support
Ventilation*
Michael j. Banner, PhD; Robert R. Kirby, MD ;
Orlando C. Kirton, MD, FCCP; C. Bryan DeHaven, BA, RRT; and
Paul B. Blanch, BA, RRT
Objective: To e valuate the relationships between di-
rectly m easured work of breathing (WOB) and va ri -
ables of the breathing pattern commonly used at the
b edside to infer WOB for intubated, sp ontane ous ly
breathing patients treated with pressure supp o rt ven-
tilation (PSV).
Design: In vivo measurements of the WOB were
obtained on a consecutive series of adults. Breathing
frequency (f), tidal volume (VT), the index of rapid,
shallow breathing (fIV T), the duration of r e spiratory
muscle contraction expressed as the ratio of inspira-
tory time over total respiratory cycle time (TIITToT)~
and a breathing pattern sco re (applied to approxi-
mately 50% of the patients) which ranks f, VT, ste rno-
cleidomastoid muscle activity, substernal retraction,
and abdominal paradox on a scale were variables ofthe
breathing pattern were also measured. The greater
the breathing pattern score, the lower the WOB and
vice versa.
Setting: Surgical ICUs in two university teaching hos-
pitals.
Patients: Sixty-seven adults (42 men and 25 women,
aged 20 to 78 yea rs ) who had acute respiratory failure
from various etiologies were studied. All patients were
breathing spontaneously receiving continuous positive
airway pressure and PSV.
Interventions: Intraesophageal pressure (indirect mea-
surement of intrapleural pressure) was measured with
an esophageal balloon integrated into a nasogastric
tube. VT was obtained by positioning a flow sen sor
between the "Y" pieee of breathing circuit and the
e nd otr ach e al tube. Data from these measurements
were directed to a bedside respiratory monitor (Bi-
core; Allied Healthcare Products; Riverside, Calif) that
calculates WOB u sing the CampbeU diagram. Patients
received PSV at levels d eemed reasonable to unload
the respiratory muscles. All measurements were ob-
tained after 15 to 20 min at each le vel ofPSV, averaged
over I min, and then variable s of the breathing pat-
'From the De pa rtments of AnestheSiology (Drs. Bann er and Kirby)
an d Physiology (D r. Bann er), University of F lorida College of
Medicine , ana Dep art ment of Respiratory Care (Mr. Blanch),
Shan ds Hospi tal, Gainesvi lle, F la, ana the Dep art men t of Surg ery,
D ivision of Surgical Inte nsive Ca re, (D r. Kirton and Mr. De-
Haven), Universi ty of Miami Schoo l of Medicine.
Th is paper was presented in part at the ann ual meeting of the
Amencan Society of Anesth eslologists, October 12, 199:f, Wash -
ington, DC .
Man usc ript received November 3, 1994; revision accepted May 5,
1995.
1338
tern were regressed with directly measured val ues for
WOB.
Results: All breathing pattern vari ables poorly pre-
dicted WOB as evidenced b y the low val ue s for thc
coefficients ofdetermination (~). Breathing frequency
correlated positively with WOB (r=0.47, p<O.OOI) and
predicted or explained only 22 % (~=.22) of thc vari-
ance in WOB. VT correlatcd ncgatively and flVT and
TIITToT cach correlated positively with WOB. How-
ever, these variables predicted only 20 to 27 % of the
variance in WOB. The breathing pattern score corre-
latcd negatively with WOB and predicted only 43% of
the variance in WOB. A prediction model taking aU
va ri ab les into consideration using multiple r egression
analysis p redicted only 50% of the variance in WOB;
thus, it too was a poor to moderate predictor ofWOB.
Conclusion: Our data reveal that WOB should be
measured directly because variables of the breathing
pattern commonly used at the bedside appear to be
inaccurate and misleading inferences of the WOB. The
clinical implication of the se findings involve s the tra-
ditional and empirical practice of titrating PSV based
on the breathing pattern. Wc do not imply that the
patient's breathing pattern should b e ignorcd, nor
undcnnine its importance, for it provides useful diag-
nostic information. It appears, however, that rclying
primarily on the breathing pattern alone does not
provide enough infonnation to accurately assess the
respiratory muscle workload. Using the breathing
pattern as the primary guideline for sele cting a level
of PSV may result in inappropriate respiratory muscle
workloads. A more comprehensive strategy is to em-
ploy WOB m easurements and the breathing pattern in
a complementary manner when titrating PSV in criti-
cally ill patients. (CHEST 1995; 108:1338-44)
CPAP=continuous positive ainvay pressure ; febreathing
frequency; flVT=ratio of brcathing frequency over tidal
volume; PSV=pressurc support ventilation; TIITTOT=ratio
of inspiratory time over total respiratory cycle time;
Vretidal volumc; WOB=work of breathing
Ke y words: breathing freq uency; breathing pattern; chest
wall compliance; esophageal pr essure; mechanical ventila-
tion; pressure support venti lation; respiratory failure; respi-
ratory monitor; respiratory muscle loading; work of breath-
ing
Clinical Investigationsin Critical Care
 Spontaneous breathing frequency (f) and a patient's
breathing pattern in general are comm only used to
infer the work of breathing (WOB) or respirato ry
muscle afterload during ventilatory support.1 A normal
or tolerable WOB is inferred when f is approximately
15 to 25 breath s/min for adults'' and accessory respi-
ratory muscles, such as the sternocleidomastoid mus-
cles, are not contracting.' It is recommended that f
should be taken into consideration when titrating
pressure support ventilation (PSV) to unload the res-
piratory muscles and decrease the WOB of patients
with increased workloads.1,2,4 For example, if the f for
an adult were approximately 35 to 40 breaths/min and
the sternocleidomastoid muscles were contracting at a
specific level of PSV, then it would be inferred that
WO B is abnormally increased and that the level of PSV
should be increased. Following an increase in PSV, if
the f decreases to approximately 15 breaths/min and
sternocleidomastoid muscle activity is absent, then it
would be inferred that the new level of PSV is appro-
priate to unload the respiratory muscles and that the
WOB is in a more tolerable range. Observing the
breathing pattern and inferrin g whether the patient's
WOB is appropriate is the conventionally used ap-
proach for selecting a clinically acceptable level ofPSV.
Ostensibly, the approach is subjective and may be
considered "educated guesswork."
The precision of using f and the breathing pattern to
infer WOB when applying PSV is in question. In sev-
eral reports , f did not correlate with WOB durin g
weaning from ventilatory support. For example, Bro-
chard et al5 reported that , following application of
decremental levels of PSV, f was fairly constant at
about 23 breath s/min, while WOB ranged from 0.55 to
1.10 J/L in patients with respiratory failure. Nathan et
al6 observed f to be fairly constant at about 22
breaths/min , while WOB ranged from 0.31 to 1.04 JIL
inlatients receiving PSV prior to extubation . Petros et
al showed that while patients breath ed spontaneously
durin g PSV, with continuous positive airway pressure
(CPAP) or with aT-piece, WOB ranged from 0.87 to
1.78 JIL, while f was fairly constant at approximately 22
breath s/min. In another study, Tokioka et alBreported
that f was fairly constant at 25 to 30 breaths/min, while
WOB ranged from 0.2 to 0.75 JIL following deere-
mental levels of Psv.
A need exists to evaluate the accuracy of predicting
WOB based on f and related assessments of the
breathin g pattern in critically ill patients who are
receiving ventilatory support. We hypothesized that
objective, real-time measurement of a patient's total
WOB (physiologic plus imposed work) is a more
accurate, and thus, appropriate, method of assessing
the respiratory muscle workload. To test this hypoth -
esis, we evaluated the relationship s between directly
measured WOB and variables of the breathing pattern
Table I- Breath ing Pattern Score (Inference of WOB)*
Variable Rank
0 2
f, breaths/min >20-30 slO-20
>30
VT, mUkg <4 4-<8 >8-10
Sternocleidomastoid Severe Mild Absent
muscle activity {noted by (noted by
observation) palpation only)
Substernal retraction Severe Mild Absent
Abdominal paradox Severe Mild Absent
*Score range: Oto 10; a score of 10 is an inference of a normal WOB
of respiratory muscle load. The lower the score, the greater the WOB
and vice versa.
commonly used at the bedside to infer WO B for intu -
bated , spontaneously breathing patients receiving PSV
who were diagnosed as having respiratory failure .
MATERIALS AND METHODS
Sixty-seven adults (42 men and 25 women) admitte d to th e sur-
gical IC U who we re diagnos ed as having acu te respirat ory failure
from various etiologies were studi ed afte r obtaining inform ed con -
sen t from the patient's family. Th e study was approved by th e In-
stitutional Review Boards at Shands Hospital at the University of
Florid a Medical Center and at the Jackson Mem orial Medical
Center at the University of Miami. Patients in the stu dy popul ation
were diagnosed as having moderate to severe forms of respirato ry
failure based on airway pressu res, pulm onary mechan ics, radiologic
examination, and blood gas exchange. Causes of respirato ry failure
included sepsis, cirrhosis, pn eumonia, congestive heart failure,
pulmonary ede ma, ascites, craniotomy, or abdomi nal and chest wall
blunt trauma from either gunshot, stab wounds, or motor vehicl e
acciden ts. A few patients had COPD . Age and weight ranged from
20 to 77 years and 40 to 180 kg, respectively. All patients received
PSV at levels deem ed reason able to unlo ad th e respiratory mu scles
and lessen th e WOB based on observing th e patients ' br eathing. All
patients had either an endotrache al or a trach eostomy tub e in place
and were breathin g spontaneously receiving CPAP from 5 to 15 cm
H20 . Endotracheal and tracheostomy tub e sizes ranged from 6.5 to
10.0 mm internal diam eter. Applied levels of C PAP were based on
established criteria of applying a nontoxic fractional inspired oxygen
conce ntration (F ,0 2 of 0.5 or less) to achieve an arterial oxygen
satura tion of 97% or more.
Sever al types of mechanical ventilators were used (H amilton
Veolar; Reno, Nev; Puritan-Bennett 7200; Ca rlsbad, Calif; Bird
6400 ST; Palm Springs, Calif; and Siemens 900 C ; Scha umburg, III).
All patients were stu died using the ventilator chose n by th e med-
ical and respiratory care pe rsonn el. Th e patient-trigger sensitivity
setti ng for all ventilators was set at - 2 em H20. Standard diam eter
corru gated tubin g and heated wick-type humidifi ers were used with
all ventilators.
At all levels of PSV, th e following breathing pattern variables
(inferences of WO B) were measured : f; tidal volum e (VT); the in-
dex of rapid shallow breathing, expresse d as th e ratio of f over VT
(/NT); th e duration of respiratory muscle contrac tion or the portion
of the respirat ory cycle when the inspiratory muscles are active,
expressed as the ratio of inspiratory time over total respirato ry cy-
cle time (T vTTOT); and a breathin g pattern score that we derived
th at ranks the following five variables each on a scale from 0 to 2:
f, VT, and th e degree of sternocleidomastoid mu scle activity, sub-
stern al retraction, and abdominal paradox (Table 1). Th e total score
ranged from 0 to 10, with 10 inferring a normal or tolerable WOB .
CHEST/ 108/ 5 / NOVEMBER, 1995 1339
 LUNGS

~ STOMACH

VENTILATOR
.- - . . J ) :~~~~5 ~
C)
' - - - 7 "."

'---------./P"'U~TAC? 1
(MEASUREMENT Of VOLUME
ANO flOW RATE)
y
1
ESOPHAGUS I NTRAESOPHAGEAL
BALLOON CATHETER
fOR MEASURI NG
ESOPHAGEAL PRESSURE

BICORE CAMPBELL DIAGRAM

~J C)iJi'W
MONITOR TOTAL WORK Of BREATHING
<r-;
IDIIIIIl RESISTIVE WORK

o + +
ESOPHAGEAL PRESSURE
~
I
ELASTIC WORK

IWORK Of BREATHING
1,5 Joule / L

FIGURE 1. Clinical meth od of measurin g the WOB performed by the patient (including physiologic plus
impo sed work). WOB is computed using the Campb ell diagram, which shows the change in volume plot-
ted over the change in esophageal pressure (inference of intrapleural pressure) durin g spontaneous inha-
lation (I) and exhalation (E). The chan ge in volume is measured at the connection between the "Y" piece
of the breathing circuit and the endotracheal tub e with a miniature pneu motachograph (flow sensor ).
Esophageal pressure is measured with an intraesophageal baUoonpositioned in the middle to lower third
of the esophagus. These data are directed to a computerized, port able monitor (Bicore, model CP-lOO).
Th e esophageal pressure-volume loop moves in a clockwise direction; the slope of the loop is lung com-
pliance (CL). Chest wall compliance (Ccw) is obtained previously by mechanically ventilating a relaxed
pat ient. Under these conditions, the esophageal pressure volume loop moves in a counterclockwise
direction (not shown); the slope of the loop is Ccw. (This compliance value is stored in the monitor's
computer memory and is used to construct the Campbell diagram.) Inspiratory resistive WOB includes
the physiologic resistive work on the airways and the resistive work on the breathing apparatus (vertical
lines). Elastic WOB is the tria ngular-shaped area subtended by the lung and Ccw curves (diagonal lines).
Total measured WOB , the sum of resistive and elastic work, is 1.5 JIL in this example.

It is inferred that the greater the breathing patt ern score, the lower
the WOB and vice versa. The selected variables and scoring range
are based on our collective experience at qu antifYingchanges in the
breathin g pattern . Th e variables used in the breathing pattern score
are commonly used by many clinicians to assess breat hing. Only 30
patients at one of the institutions (University of Florid a) wer e eval-
uated with the breathing patt ern score.
WOB performed by the patient was also directly measured at all
levels of PSV. Total measured WOB , which includ es physiologic
(elastic and resistive ) work and the imposed resistive work of the
breathing appa ratus (endotracheal tub e , breath ing circuit tub ing,
and ventilat or's demand-flow system ) during spont aneous ventila-
tion was calculated by integrating the change in intraesophageal
pressure (indirect measurement of intr apleu ral pressure ) and VT.9
Intraesophageal pressure was measured (relative to atmospheric
pressur e) by the nasal insertion of an esophageal balloon that is in-
tegrated into a nasogastric tub e (Smart Cath, Nasogastric Balloon
Cathete r; Bicore: Allied Health care Products; Riversid e, Calif).
Correct position was confirm ed with an occlusion test as described
by Baydur et al:1O following occlusion of the airway opening, the
changes in pressure at the airway and in the esophagus are close to
identical during spontaneous inspiratory efforts. VTwas obtained by
integrating the flow signal of a pneum otachograph (flow sensor )
positioned betwe en the "Y" piece of the breathing circuit and the
endotrache al tub e . Airway pressure was measured from a cathe ter
attached to the pneumotachograph ; this was used to measure CPAP
and PSV. Data from these measurements and from the measure-
ment of chest wall compliance were processed with a portabl e,
comm ercially available respiratory monitor (Bicore, model CP- 100)
that calculates WOB by the method described by Campbell' ! and
Agostoni et al12 (Fig 1). This is a well-established method that has
been used by many investigators to assess WOB during ventilatory
support.13·16 The nasogastric tub e with the esophageal balloon and
the pneum otachograph are components of the monitor , which also
provides real-time measurements of f, fi'VT, and TIffTOT . Th e de-
gree of ste rnocleidomastoid muscle activity, substernal retraction,
and abdominal paradox were ranked based on clinical presentation,
ie, observati on and palpation by a board-certified anesthesiologist
who was an attendin g physician in critical care medicine.
Accuracy in measuring chest wall compliance requires a relaxed
and mech anically ventilated patient. Patients were initially given 1
to 2 mg of midazolam for relaxation and then the mechani cal ven-
tilator rate was increased transiently to about 12 breaths/min. Un-
der these condition s of mechanical inflation with a preselected VT
esophageal pressure increases. The monitor integrates the changes
in esophageal pressure and volume to produce a pressure-volume
loop moving in a counterclockwise direction, the slope of which was
taken as the chest wall compliance. This compliance value was
stored in the monitor's compute r memo ry. Measured chest wall
compliance values were 109± 37 mUcm H20. Next, 0.2 mg of flu-

1340 Clinical Invesugalions in Critical Care

mazenil was given to reverse the effects of the midazolam so that Table 2-Relntionships Between Directly Measured
the patient resumed spontaneous breathin g. Th e dose was rep eated W OB and Variables Used to Pred ict or Infer WOB
if complete reversal of sedation was not achieved in 10 min. The
mechanical ventilato r rate was decreased to zero and PSV reapplied. Relationship to Measured WOB
T he monitor places the static chest wall comp liance line at the 0
deflection in baseline airway pressure that marks the onset of each Predictor Variables r p
spontaneous inhalation. j orm ally, this deflection and the start of f 0.47 0.22 <0.001
inspiratory flow coincide. The developm ent of intrinsic positive VT - 0.50 0.25 <0.001
end-expiratory pressure, however, produces a threshold load re- Index of rapid, shallow 0.45 0.20 <0.05
qu iring the respiratory muscles to generate sufficient force to breathing (fNT)
counterac t the oppoSing positive recoil pressure before flow Duration of respiratory 0.52 0.27 <0.006
begtns.!? Under these conditions, the static chest wall compliance muscle contraction,
line is placed before flow comm en ces and is displaced horizontally expressed as TIff TOT
from the zero flow point and the lung compliance line. Separation Breathing pattern score -0.65 0.43 <0.001
of the two compliance lines increases the area enclosed within the
Ca mpbell diagram and, in tum, increases elastic and thu s, total
WOB . At the onset of each spontaneous inhalation, the monitor
measures intrin sic positive end-expiratory pressure as the change in 2. 5
intra esophageal pressure required to effect a change in the baseline
Ir2 = .2 2 1 •
airway pressure .
• •
......
WO B perform ed by the ventil ato r to inflate the respiratory sys- 2.0
•
• • • •• ••
tem du ring PSV, which was not measured in this study, may be
calculated by the monitor; pressure measured at the airway open - Cl 1.5 • '

ing is integrated with VTto produce a pressure-volume loop . The c

area circumscribed within this loop is defined as the work per-
:c.... •
'

CO 1.0
formed by the ventilator. Q) Q)
•
•
.... 0>

it
In a validation study of the accuracy of the monitor, 18 WOB was CO c
6
measured by using the monitor and simultaneously using conven- '0
~ E 0. 3
tional equipment to construct a Ca mpbell diagram. The relationship .... 0
between the two sets of measurem ents of WO B was nearly perfect o z
(r'!=0.99, p <O.oo I) . Bias was minimal (- 0.05 JIL ) and precision was ~ o 5 10 15 20 25 30 35 40
excellent (+0.03 JIL ). '------..y----'
All measu rement s (over 200 detenninations) were obtained af- Clinically acceptable range

ter IS to 20 min at each level of PSV, 5 to 10 pe r test, averaged over
1 min, and then reco rded. Measurements were record ed only if they
eould be consistently rep rod uced during each set of tests; thus, in-
tern al test eonsistency (reliability of measurem ent) was ensure d.
When appropriate, arte rial blood gas and hemodynamic data were
co llected at various levels of PSV. All variables or inferences of
WOB were regressed with directly measured values for WOB at all
levels of PSV. Data were analyzed by linear, polynomial (seeond
order), and multiple regre ssion analyses. Alpha was set at 0.05 for
statistical significance.
Spo ntaneo us Breat hing Frequency (per m in )
FIG URF: 2. Helationship be twee n \ VO B and spontaneous f is de-
picted. Although a positive and significant eorre lation was found
(r=0.47, p <O.OOl ), spo ntaneous f pred icted or explained only 22%
of the variance in WOB , ie, a poor predictor. With in a clinically
accep table f range of 15 to 25 breaths/min, some patient s performed
no work (predisposing to disuse respira tory muscle atrophy), oth-
ers performed work in a norm al range, and some perform ed work
in a fatiguing range (predisposing to resp iratory muscle fatigue ).

R ESULTS
Measured values for WOB ranged from 0 to 2.2 JIL.
The level of PSV applied ranged from 5 to 50 em H 20 .
All breathing pattern variables poorly predicted WOB
as evidenced by the low values for the coefficients of
determination ( ~). For each variable, ~ predicts or
explains the amoun t of variance in WOB. A variable
with an ~ value between 0.64 and 0.81 is considered
high and thus, a fairly good predictor. 19 The f corre-
lated positively with WOB and predicted only 22% of
the variance in WOB (Table 2 and Fig 2). VT correlated
negatively and for fNT and TIffTOT each correlated
positively with WOB. However, these variables pre-
dicted 20 to 27% of the variance in WOB (Table 2).
The breathing patt ern score correlated negatively with
WOB and it predicted only 43% of the variance in
WOB (Table 2, Fig 3).
By using multiple regression analysis, a prediction
model was formulated that took all variables into con-
sideration . Only three variables were significant pre-
dictors ofWOB: breathing pattern score, TIffTOT, and
VT. However, because it predicted only 50% of the
variance in WOB, the model was a poor to moderate
predictor of WOB (Table 3).
For most ventilatory conditions, all patients were in
hemodynamically stable condition and blood gas ex-
change was within acceptable ranges.
DISC USSIO;o.J
The main finding of this study is that, for adults with
abnormal pulmonary mechanics and loaded respira-
tory muscles who are in respiratory failure and being
treated with PSV, the f and the breathing pattern are
poor inferences of the WOE. Our study reveals that
WOB should be measured directly because f, for ex-
ample, appears to be an inaccurate and misleading
variable from which to infer the respiratory muscle
workload. The clinical implication of these findings

CHEST / 108 / 5/ NOVEMBER, 1995 1341

 I
Tab le 3-Predic tion (Multiple Regression)
2
2 .5
r = .43 Mode l f or WOB

-l • •
(r2=O.50, p<O.OOl )

• •
...... WOB=2.14+0.0002 (D- 0.44 (VT)*- O.OO3 (fN T)
2 +2.13 (TuTToT)*- 0.21 (breathing pattern score)"
Q)

•• ••••
::J
0 *VT, the duration of respiratory muscle contraction expressed as
:3 TuTTOT, and the breathing pattern score were the only significant
1 .5
(p<0.05) predictors (regressors) in the model, where f is breathing
• •
C)
c
•
frequen cy (breaths/min) and fNT is the index of rapid, shallow
x:
......
til • breathing (breaths/rnin-L). The factor "Institution" (University of
Florida and University of Miami) is not shown because it also was not
•
Q)
L-
eo a Significant regressor. The model predicts or explains Dilly 50% of the
......
0
.5
• • variance in WOB, ie, a relatively poor to moderate prediction model.

sc
L-
0
3 dex of rapid shallow br eathin g does not app ear specific
0 enough to infer WOB when applying PSV. Moreover ,
0 2 4 6 8 10
BreathIng Patte rn Scor e
FIGURE 3. Relationship between WOB and th e breathin g pattern
score (Ta ble 1) is shown . WOB correlated negatively and signifi-
cantly with the breathin g pattern score (r=- 0.65, p <O.OO l) . H ow-
ever, the breathing patte rn score was a poor pred ictor explaining
only 43% of the variance in WOB. ideally, a breath ing pattern score
of 10 is an inference of a normal wor kload (0.3 to 0.6 JIL). How-
eve r, with this score, some patients pe rforme d no work, others
pe rforme d work in a norm al range, and some perform ed work in
a potentially fatiguing range.
brings into qu estion the traditional practice of titrat ing
PSV based on breathing pattern, as this may resu lt in
inappropriate respiratory muscle workloads. This point
is illustrated in Figure 2. Whil e receiving PSV and
breathing at a clinically acceptable range between 15
and 25 breaths/min, the WOB for some pati ents was
eithe r zero, in a normal range (0.3 to 0.6 J/L 4,20,21), or
at an abnormally high rang e. Similar findin gs were
observed with th e breathin g pattern score (Fig 3).
Disuse respiratory muscle atrophy may result when the
muscles are totally un loaded and performing no work
for an exte nde d period. 22 Respiratory muscle fatigue is
likely to resu lt when the muscles are forced to perform
workloads at an abnormally high rang e for too long a
period. With eithe r severe respiratory muscle atrophy
or fatigue, pati ents may becom e dependent and not
wean from ventilatory support.
Our findin gs are in accordance with oth er studies .
Silas et al23 reported that the index of rapid shallow
breathing (fNT) as described by Tobin et al24 was a
poor predictor ofWOB because it predicted only 16%
of the varianc e in WOB in patients before extu bation.
The ind ex of rapid shallow br eathing predicted only
20% of th e variance in WOB in our patients. Most of
our patients were not ready for weaning from ventila-
tory support but were treated during the initial-to-late
ph ases of respiratory failure . Althou gh regard ed as an
inference of the respiratory muscle workload, the in-
regarded to be an accurate predictor of weaning fail-
ur e, the index has also been criticized as poorly
predicting extubation outcorne.P Th e ind ex lacked
sufficient sen sitivity and speci ficity to differen tiate
between patients who were successfully extubated and
tho se who we re not.
Because it predicted only 50% of the variance in
WOB , the prediction model is not considered a clin-
ically acceptable method of inferring WOB . Other
factors explaining th e variance in WOB not included in
the mod el may be carbon dioxide minute production ,
oxygen consumption, lung and ches t wall compliances,
airway resistance, physiologic dead space, and peak
inspiratory flow rate demand. A mor e comprehensive
prediction mod el, including the aforem entioned vari-
ables, may increase the value of r2 and , thus , its pre-
dictive ability. However , it is easier to measur e WOB
dir ectly than to measure and regress a number of
physiologic variables. Predicting WOB based on a host
of clinical variables, even assuming that all measure-
ments are obtained correctly, appears dicey.
We do not dispute the fact that as pu lmonary me -
chanics det eriorate, th e respiratory muscles are loaded,
WOB increases , and the br eathing pattern changes.
Specifically, these changes are vagally mediated by af-
ferent or sen sory fibers (load sensors) in the lungs and
respiratory tract. Three types of affere nt fibers that
modulate the breathing pattern have been describ ed :
(1) slowly adapting receptors; (2) rapidly adapting re-
cep tors (also termed "deflation," "cough," or "irritant"
receptors ), both of which are pulmonary stretch or
mechanoreceptors; and (3) chemos ensitive or C-fibe r
endings.26 Slowly adapting receptors are found in the
bronchial smooth muscle fibers, rapidly adap ting re-
cep tors in the supe rficial layers of th e respiratory tract
mucosa, and C-fibers in the airway epithelium .27 These
recep tors det ect changes in pulmonary mechanics and
thoracic gas volume (functional residual capacity).27,28
Fo llowing a decrease in lung compliance (increase
in respiratory muscle load), an increase in discharge

1342 Clinical Investigations in CriticalCare

activity from the stretch receptors increases. Similar
responses follow increases in total resistance. 27,28 C-
fiber endings are activated by many substances pro-
duced in the lungs , such as histamine, bradykinin, and
some prostaglandins.27,28 Some sympathetic afferents
may also be activated by increases in mechanical
loads.27 Afferent discharge Signals from the sensory
fibers are directed back via the vagus nerve to the
central respiratory controllers in the central nervous
system, modifying their output Signal, which in tum,
modifies the breathing pattern.P' Stimulation of these
receptors produces patterns of rapid, shallow breath-
ing to minimize large changes in intrapleural pressure.
Patients with loaded respiratory muscles breathe at a
faster rate and a smaller VT to minimize the WOB-
the so-called minimal WOB or least average force
concept, which produces the most energy-efficient
combination ofVr and [,20,30,31 This f may be described
as optimal because, when the frequency is too low,
much elastic work is required to produce large VT;
when f is too high , much resistive work is required, as
well as work uselessly done in ventilating the dead
space with each breath.i" This mechanism also func-
tions to protect th e respiratory muscles from exhaus-
tive, fatiguing contractions that could lead to muscle
fiber splitting, hemorrhage, and self-destruction.P'
A local-load compensator mechanism involving
muscle spindle receptors and motor neurons in the
intercostal muscles and diaphragm has also been
described.27 This mechanism regulates muscle con-
traction to obtain the "desired" change in length, and,
thus, VT. The demand for a given change in length is
transmitted from the muscle receptors to the motor
neurons. As a result, a change in mechanical load (eg,
increased airway resistance) to the contracting muscle
leads to compensatory adjustments that increase the
activity of the motor neurons ,
Unquestionably, this is a sophisticated, physiologic,
load-sensing, and load-compensating mechanism that
functions during spontaneous breathing. Whether this
mechanism functions in the same manner during
assisted spontaneous breathing with PSV is unknown.
However, some altered role of the pulmonary stretch
receptors in mediating the response seems likely, par-
ticularly when ventilatory support results in larger than
normal VT(this is often the case during PSV) and res-
toration of functional residual capacity by applying
CPAP. The focal point or question of this argument is,
can the aforementioned physiologic, load-sensing, and
load-compensating mechanism be relied on alone to
predict or infer respiratory muscle workloads accu-
rately for all patients in res~iratory failure treated with
PSV? Our study and others -8,23 reveal that the breath-
ing pattern is not an accurate predictor ofWOB and
may provide preliminary evidence that this physiologic
mechanism is indeed affected during positive pressure
ventilatory support. There is also evidence that the
perceived inspiratory effort sensation during sponta-
neous breathing (how the patient feels, degree of
comfort) is not related to fatiguing or nonfatiguing di-
aphragmatic contractions.V Thus, how a patient "feels"
also may not be a reliable inference of respiratory
muscle workloads.
A logical deduction is that WOB should be mea-
sured directly. With easy-to-operate, portable bedside
respiratory monitors, real-time measurements ofWOB
are readily obtainable." Directly measured values for
WOB may serve as objective and quantifiable data for
applying PSV to achieve specific goals, ie, to partially
or totally unload the respiratory muscles. P A total
WOB of 2 JIL (approximately 300% above normal)
while a patient receives 10 cm H 20 PSV, for example,
would be considered by many clinicians to be a
potentially fatiguing workload. PSV can be increased
until the respiratory muscles are partially unloaded and
the WOB is in a more tolerable or perhaps normal
range. Totally unloading fatigued respiratory muscles
and allowing them to rest and recover is appropriate
under some circumstances.34-36 Muscle fatigue is re-
versible with rest and is the most important way to treat
it.34 The time for respiratory muscles to recover from
chronic fatigue is estimated to be at least 24 h. 34,35
Some patients with COPD and respiratory muscle fa-
tigue may benefit by decreasing the WOB to zero and
totally unload the muscles for approximately 24 h .
Subsequently, PSV may be decreased so that the WOB
is in a more normal range and the respiratory muscles
are partially unloaded. The load tolerance for each
patient may be accurately and objectively determined
rather than subjectively inferred (guessed) by observ-
ing the breathing pattern only.
In our experience over a 3-year period, directly
measured WOB data were useful for selecting appro-
priate levels of PSV for approximately 50% of our pa-
tients. These data were particularly helpful when we
were uncertain whether the respiratory muscles were
properly unloaded for patients with moderate-to-
advanced respiratory failure during weaning from
ventilatory support when measurements of imposed'"
and physiologic WOB were useful in identifyinf- those
patients who could be successfully extubated.' -40 Pa-
tients with milder forms of respiratory failure usually
do not need this type of monitoring. In others,
however, the measurements appear essential to guide
therapy appropriately.
The study was specifically designed not to involve a
homogeneous subset of patients and ventilatory equip-
ment. Conclusions and clinical implications of such a
study would be limited in scope. Rather, a more het-
erogeneous study involving patients with a variety of
diagnoses and using various types of ventilators allows
more general conclusions to be drawn.
CHEST / 108 / 5 / NOVEMBER , 1995 1343
 In summary, we contend that the traditional and
somewhat subjective practice of inferring a level of
WOB or respiratory muscle afterload during PSV
based on bedside assessments of the breathing pattern
alone appears inaccurate. We do not undermine the
importance of assessing a patient's breathing pattern
and do not imply that it should be ignored. On the
contrary, the breathing pattern provides useful diag-
nostic information and should be followed in patients
receiving ventilatory support. Changes in the f and
pattern are the result of changes in the workload per-
formed by the respiratory muscles. However, we have
shown that relying primarily on the breathing pattern
alone does not provide enough information to accu-
rately assess the patient's WOB when titrating PSV.
WOB measurements and breathing pattern should be
used in a complemen tary manner when titrating PSV
in critically ill patients.
R EFERENCES
Macintyre NR. We aning from mechanical ventilato ry suppo rt:
volume-assisting intermittent breath s versus pressure-assisting
every breath . Respir Care 1988; 33:121-25
2 Kacmarek RM. Point of view: pressure suppo rt. Respir Care
1989; 34:136-38
3 Macintyre NR, Nishimura M, Usada Y, et al. Th e Nogoya con-
fer en ce on system design and patient-ventil ator interactions
du ring p ressure support ventilation. Chest 1990; 97:1463-66
4 Kacmarek RM . Th e role of pressure support ventilation in
reducing work of breathing. Resp ir Care 1988; 88:99-120
5 Broch ard L, Rua F , Lorin o H, et al. Inspiratory pressure support
compensates for the additional work of breathing caused by the
endo tracheal tube . Anesth esiology 1991; 75:739-45
6 Nathan SD , Ishaaya AM, Koerner SK, et al. Predicti on of min -
imal p ressure support during weaning from mechanical ventila-
tion . Chest 1993; 103:1215-19
7 Petros AJ, Lamond CT , Bennett D. Th e Bicore pulmonary
monitor. Anaesth esia 1993; 48:985-88
8 Tokioka H, Kinjo M, Hirakawa M. Th e effectiveness of pressure
support ventilation for mec hanical ventilatory suppo rt in children .
Anesth esiology 1993; 78:880-84
9 Banner MJ, Jaeger M, Kirby RR. Compone nts of the work of
breathin g and implications for monitoring ventilator-dependent
patient s. C rit Care Med 1994; 22:515-23
10 Baydur A, Behrakis P, Zin WA, et al. A simple method for
assessing the validity of the esophageal balloon technique. Am
Rev Respir Dis 1982; 126:788-91
11 Campbell EJM . Th e respiratory muscles and the mechanics of
breathin g. Chicago: Year Book Publishers , 1958; 1-117
12 Agoston i E, Ca mpbe ll EJM , Freedman S. Ene rgetics. In: Camp-
bell EJM , Agostoni E . Davis IN , cds. Th e respiratory muscles:
mechanics and neural control. Philad elph ia: WB Saunde rs, 1970 ;
115-20
13 Brochard L, Rua F , Lorino H, et al. Inspiratory pressure support
compensates for the additional work of breathing cause d by the
en dotrac heal tu be. Anesth esiology 1991; 75:739-45
14 Brochard L, HarfA, Lorino H, et aI. Inspiratory pressure support
prevents diaphragmati c fatigue during weaning from mech anical
ventilation. Am Rev Respir Dis 1989; 139:513-21
15 Sassoon CS, Giron AE , Ely EA, et aI. Inspiratory work of
breath ing on flow-by and dem and-flow continuous positive airway
pressure. C rit Care Med 1989; 17:1108-14
1344
16 Van de GraaffWB, Gordey K, Dornseif SE , et aI. Pressure sup-
port : changes in ventilatory patt ern and components of the work
of breathin g. Ch est 1991; 100:1082-89
17 Tobin MJ, Lodato RE. PEEP, auto- PE EP, and waterfalls. Ch est
1989; 96:449-51
18 Blanch PB, Bann er MJ. A new respiratory monitor that e nables
accurate measurement of work of breathin g: a validation study.
Respir Care 1994; 39:897-905
19 Hu ck SW, Coo nnier WH, Bounds WG. Readin g statistics and
research . New York: Harper & Row, 1974; 148-76
20 Otis AB. The work of breathin g. In: Fen n W 0 , Rahm H , eds.
Hand book of physiology: a critical, comp re hensive p resentation
of physiological knowledge and concepts; section 3: respiration. I-l
Washingto n, DC : American Physiological Society, 1964; 463-476
21 Amato MBP , Barbas CSV, Bonassa J, et aI. Volum e assure d
pressure suppo rt ventilation (VASpV): a new approach for
redu cing muscle workload durin g acute respiratory failure. Ch est
1992; 102:1225-34
22 Marini JJ. Exertion during ventilatory suppo rt: how much and
how important. Hespir Care 1986; 31:385-87
23 Silas SL, Simpson SQ, Levy H . Rapid shallow breathing index
does not correlate with airway occlusion pressur e or work of
breathin g [abstract]. Chest 1993; 104:130S.
24 Tobin MJ, Skorodin M, Alex CG . Weaning from mechanical
ventilation. In : Taylor RW, Shoemake r WC , eds. Critical care
state of the art . Fu llerton, Calif: Society Critical Care Medicin e,
1991: 373-413
25 Lee KH, Hu i KP, C han TB , et al. Rapid shallow breathing (fre-
qu en cy-tidal volume ratio ) did not p redict extubation outcome.
Chest 1994; 105:540-43
26 Barnes PJ. Ne ural control of airway smoot h muscle. In: Crystal
RG,West JB, eds.Th e lung. New York: Raven Press, 1991; 903-16
27 Sant' Ambrogio G, Sant' Ambrogio F B. Reflexes from the airway,
lung, chest wall, and limbs. In: Crystal RG, West JB, eds , Th e
lung. New York: Raven Press, 1991; 1383-95
28 Sant' Ambrogio G. Information arising from the tracheo bronchial
tree of mammals . Physiol Rev 1982; 62:531-69
29 Roussos C. Function and fatigue of respiratory mu scles. C hest
1985; 88(suppl):I24 S-32S
30 Otis AB, F enn WO , Rahn HL. Mechanics of breathi ng in man .
J Appl Physiol 1950; 2:592-607
3 1 Kacmarek RM , Ven egas J. Mechanical ventil atory rates and tidal
volumes. Hespir Care 1987; 32:466-75
32 Bradley TO, C hart rand DA, Fitting JW, et aI. Th e relation of in-
spiratory effort sensation to fatiguing patterns of the diaphragm.
Am Rev Hespir Dis 1986; 134:1119-24
33 Banner MJ, Kirby RR, Gabrielli A, et al. Partially and totally un-
loadin g resp iratory muscles based on real time measurem ent s of
work of breathing: a clinical app roach . C hest 1994; 106:1835-42
34 Grassino A, Macklem PT. Respiratory muscle fatigue and venti-
latory failure. Ann Rev Med 1984; 35:625-47
35 Braun NMT, Faulkner J, Hu ghes RL, et aI. Wh en should respi-
rato ry muscles be exercised? Chest 1983; 84:76-84
36 Stoller JK. Physiologic rationale for resting the ventilatory mus-
cles. Respir Care 1991; 36:290-96
37 Bann er MJ, Kirby RR, Blanch PB, et aI. Decreasing imposed
work of the breathing appara tus to zero using pressure suppo rt
ventilation. Crit Care Med 1993; 21:1333-38
38 Banner MJ, Gabrielli A, Kirby RR, et aI. Extubating at a pressure
suppo rt ventilation level corresponding to zero imposed work of
breathi ng [abstract]. Anesthesiology 1994; 81:A271
39 Kirton OC, DeHaven B, Morgan J, et aI. Endotracheal tub e flow
resistance and elevated impo sed work of breathin g masquerading
as ventilator weaning intoleran ce [abstract] . Chest 1993; 104:133S
40 Kirton OC, DeHaven B, Morgan J, e t aI. Work of breathin g
analysis during tachypn eic extubation trial failures reduces false
negative classification [abstract]. Chest 1994; 106:159S
Clinical Investigationsin Critical care