Patient-ventilator pressure-support

interaction during vs. proportional-assist

acute hypercapnia: ventilation

V. MARCO RANIERI, ROCCO GIULIANI, LUCIANA MASCIA, SALVATORE GRASSO, VITO PETRUZZELLI, NUCCIA PUNTILLO, GAETANO PERCHIAZZI, TOMMASO FIORE, AND ANTONIO BRIENZA Istituto di Anestesiologia e Rianimaxione, Ospedale Policlinico, Universith di Bari, 70100 Bari, Italy
Ranieri, V. Marco, Rocco Giuliani, Luciana Mascia, Salvatore Grasso, Vito Petruzzelli, Nuccia Puntillo, Gaetano Perchiazzi, Tommaso Fiore, and Antonio Brienza. Patient-ventilator interaction during acute hypercapnia: pressure-support vs. proportional-assist ventilation. J. Appl. Physiol. 81(l): 426-436, 1996.-The objective of this study was to compare patient-ventilator interaction during pressure-support ventilation (PSV) and proportional-assist ventilation (PAV) in the course of increased ventilatory requirement obtained by adding a dead space in 12 patients on weaning from mechanical ventilation. With PSV, the level of unloading was provided by setting the inspiratory pressure at 20 and 10 cmHzO, whereas with PAV the level of unloading was at 80 and 40% of the elastic and resistive load. Hypercapnia increased (P < 0.001) tidal swing of esophageal pressure and pressure-time product per breath at both levels of PSV and PAV During PSV, application of dead space increased ventilation (VE) during PSV (67 t 4 and 145 t 5% during 20 and 10 cmHsO PSV, respectively, P < 0.001). This was due to a relevant increase in respiratory rate (48 t 4 and 103 t 5% during 20 and 10 cmHzO PSV, respectively, P < O.OOl>, whereas the increase in tidal volume (VT) played a small role (13 t 1 and 21 t_ 2% during 20 and 10 cmHzO PSV, respectively, P < 0.001). With PAV, the increase in irE consequent to hypercapnia (27 IL 3 and 64 t 4% during 80 and 40% PAV, respectively, P < 0.001) was related to the increase in VT (32 t 1 and 66 t 2% during 80 and 40% PAV, respectively, P < O.OOl), respiratory rate remaining unchanged. The increase in pressure-time product per minute and per liter consequent to acute hypercapnia and the sense of breathlessness were significantly (P < 0.001) higher during PSV than during PAY Our data show that, after hypercapnic stimulation of the respiratory drive, the capability to increase VE through changes in VT modulated by variations in inspiratory muscle effort is preserved only during PAV, the compensatory strategy used to increase VE during PSV requires greater muscle effort and causes more pronounced patient discomfort than during PAV carbon dioxide stimulation; weaning

ASANALTERNATIVETOSTANDARD

volume-controlledmodes of mechanical ventilation, considerable interest has recently developed in partial ventilatory-support techniques. Their purpose is to assist each spontaneous breath by providing positive pressure at the airway opening (Pao) during spontaneous inspiration. Pressuresupport ventilation (PSV) is the standard form of partial ventilatory support that can be used in spontaneously breathing patients, both in the early phase of acute respiratory failure (ARF) or during weaning from mechanical ventilation (6, 7, 17). During PSV, each
426 0161-7567/96 $5.00 Copyright

spontaneous breath is assisted by a constant positive pressure applied in the ventilator circuit during the duration of inspiration. Breathing frequency is determined by the patient because no controlled cycle is delivered. Tidal volume (VT) depends on the combined action of the pressure generated by the respiratory muscles (Pmus), the ventilator (Pappl), and the total impedance of the respiratory system (6, 7, 17). Proportional-assist ventilation (PAV) is an alternative mode of partial ventilatory support recently proposed (32, 35), in which the ventilator generates pressure in proportion to patient effort; the more the patient pulls, the more pressure the machine generates. The ventilator amplifies patient effort without imposing any ventilatory or pressure targets. The main theoretical difference between PSV and PAV is the function that Pappl is designed to follow during inspiration (33). With PSV, the ventilator causes Pappl to rise to a preset level and remain at that level until the cycle-off criterion is reached. The amount of Pappl should not therefore be affected by the effort of the patient (6, 7, 17). By contrast, with PAV, the time course of Pappl is linked to the time course of the effort of the patient. Pappl rises as long as inspiratory-muscle effort is produced by the patient. What is preset is not a target pressure but the proportion between Pao and inspiratory-muscle effort, i.e., how much Pappl will rise for a given increase in Pmus during inspiration (32,35). On the basis of these theoretical distinctions, substantial differences in patient-ventilator interaction are expected to occur between PSV and PAV during variations in ventilatory requirement. The aim of this study was therefore to assessthese differences and evaluate their physiological relevance in mechanically ventilated patients during the weaning period during acute variations in ventilatory requirement. We investigated the hypothesis that, whereas during PSV the increase in minute ventilation (VE) consequent to hypercapnic chemoreceptor stimulation would primarily be accomplished by an increase in respiratory rate, with PAV this would occur through an increase in VT with few effects on respiratory rate. We expected this compensatory strategy to result in greater patient comfort and lower work of breathing.
METHODS

Twelve patients admitted to the intensive care unit of the Policlinico Hospital (University of Bari) were studied. They were nasotracheally intubated (Portex cuffed endotracheal tube with an inner diameter varying from 7 to 8 mm) and mechanically ventilated for the management of ARF. EtioloPhysiological Society

o 1996 the American

EFFECTS

OF CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

427

gies of ARF included multiple trauma (5), myocardial infarction (3), pneumonia (2), and sepsis (2). Entry criteria included clinical and hemodynamic stability and a maximal inspiratory pressure of at least -20 cmHsO. During the study, no positive end-expiratory pressure (PEEP) was applied. All patients were in the weaning process, as prescribed by their attending physician. The investigative protocol was approved by the local ethics committee, and written informed consent was obtained from each patient. A physician not involved in the study protocol was always present to provide for patient care. Flow rate (V) was measured with a heated pneumotachograph (Fleisch no. 2; Fleisch, Lausanne, Switzerland) connected to a differential pressure transducer (MP-45, t2 cmHzO; Validyne, Northridge, CA), which was inserted between the Y piece of the ventilator circuit and the endotracheal tube. The pneumotachograph was linear over the experimental range of V. Equipment dead space (not including the endotracheal tube) was 70 ml. Pao was measured proximal to the endotracheal tube with a pressure transducer (Validyne, MP-45, 5100 cmH20). Changes in pleural pressure were estimated from changes in esophageal pressure (Pes) measured by using a double-lumen nasogastric tube (1.4-Fr, 127 cm; Mallinckrodt, Argyle, NY) with a thin-walled vinyl balloon (10 cm long, 3.8 cm circumference) incorporated in the lower midportion of the tube and connected to a differential pressure transducer (Validyne, MP-45, t 100 cmHz0). The esophageal balloon was filled with l-l.5 ml of air and positioned as previously described (2). All the above variables were displayed on an eight-channel strip-chart recorder (7718A, Hewlett-Packard, Cupertino, CA) and collected on a personal computer through a 12-bit analog-todigital converter at a sample rate of 100 Hz. Subsequent data analysis was performed by using the software package Anadat (RHT-InfoDat, Montreal, Quebec). VT was computed by the digital integration of the V signal. End-expiratory CO2 tension (PETIT > was continuously monitored between the pneumotachograph and the endotracheal tube with an infrared side-stream capnograph (Capnomac Ultima, Datex, Helsinki, Finland). A 20-gauge Teflon catheter was inserted percutaneously for arterial blood sampling and measurement of blood gases (IL 1303, Instrumentation Laboratories, Lexington, MA). Patients were studied in the Experimental procedure. semirecumbent position. Maximal inspiratory pressure was measured as the most negative airway pressure generated within the first 20 s of total airway occlusion at end-expiratory lung volume (19). A one-way valve was used in the expiratory line to ensure that efforts began from a lung volume less than or equal to the equilibrium volume of the chest. All patients were responsive and were instructed to make forceful efforts during this manuever; they did not usually need the full allotted time to reach a peak value. Before the study, all patients were on PSV (Siemens Servo Ventilator 9OOC, Siemens Elema, Berlin, Germany) ranging between 10 and 18 cmHZO [12.90 t 0.87 (SE) cmHaO]. Static elastance (Est,rs) and total resistance of the respiratory system (RT,rs) were obtained by using the rapid airwayocclusion technique (10) during a period of respiratory muscle inactivity obtained by injection of a short-acting hypnotic agent (propofol, 0.3 mgekg-l=rnin-l for 5 min) and controlled mechanical ventilation (VT of 10 ml/kg, constant inspiratory V of 1 l/s, and respiratory rate of 12 breaths/min). Briefly, an end-expiratory airway occlusion was performed by pressing the end-expiratory hold knob on the Siemens 9OOC ventilator. Static intrinsic PEEP (PEEPi,t) was measured as the plateau pressure during the end-expiratorv airway occlusion on the

Pao signal. At the end of the following breath, an end-inspiratory occlusion (EIO) was performed by pressing the end-inspiratory hold knob on the ventilator. After EIO, the Pao signal exhibited an initial drop [maximum pressure (P,,,) - pressure after first drop], followed by a slow decline to an apparent plateau pressure (10). The corresponding pressure value at 3-5 s after EIO was taken as the static end-inspiratory recoil pressure of the respiratory system (Pst,rs) (10). Est,rs was computed by dividing the values of (Pst,rs - PEEPist) by VT. RT,rs was calculated by dividing (P,, - Pst,rs) by V immediately preceding the occlusion (10). Twenty to thirty minutes after static respiratory mechanics measurements, when respiratory muscle activity was regained (judged to have occurred when negative swings in Pes developed during inspiration), and patients awoke, the Siemens 9OOC was replaced by a Winnipeg ventilator (University of Manitoba, Winnipeg, Manitoba). This is an experimental prototype designed to provide PAV but can also deliver all conventional modes of ventilation. The design and operation of this unit are similar to those previously described (32, 35). The gas-delivery system consists of a freely moving piston reciprocating within a chamber. The electronics control a motor that moves a piston toward the patient. As the patient breathes in, the piston moves freely into the cylinder, providing an initial flow and volume; when the velocity of the piston movement (i.e., inspiratory flow) reaches a preset threshold value, the motor starts to assist the movement of the piston. The flow level required to trigger the motor is variable, and in the present study was set at 0.05 l/s (35). When the piston is activated, it creates pressure in the piston chamber, and the forward movement of the piston produces airflow. This air is directed to the patient through a one-way valve, the inspiratory line, and the humidifier. The difference between the chamber pressure and the proximal airway pressure creates a trigger or assist signal. This signal, in turn, switches the three-way solenoid valve connecting the pump pressure to the exhalation valve line and causes the exhalation valve to close. Gas transfer is terminated when the patient stops inspiratory flow. This creates positive pressure with respect to the chamber pressure. When patient pressure exceeds chamber pressure, the trigger signal goes off, causing the exhalation valve line to open. This allows passive deflation through the exhalation valve. The piston returns to the starting position, intaking gas from the ventilator input as it moves backward. An external-demand blended-gas system is attached to the input opening. When the piston has returned to the starting position, the machine resets and is ready for the next inspiration. PEEP can be applied by adding a PEEP valve on the exhalation line of the ventilator circuit. The motor applies force to the piston according to different command signals regulated by the operating ventilatory mode. During PSV, the control denotes the level of pressure assistance for a patientinitiated breath. The changeover from inspiration to expiration can be set as different absolute flow values or as different percentage values of the initial peak flow. In this study, cycling between inspiration and expiration occurs when the flow rate falls to 5 l/min. During PAV, the command signals are instantaneous inspired flow (derived from the rate of forward motion of the piston) and instantaneous inspired volume (derived from piston displacement since the onset of inspiration). Once flow begins, pressure in the chamber rises in proportion to ongoing flow and volume, thereby augmenting the pressure gradient for chest expansion. The amount of assistance for both flow and volume can be set as varying percent values of RT,rs and Est,rs through external controls. When this percentage is set below natient resistance

428

EFFECTS

OF

CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

and elastance, chamber pressure increases only if the patient effort increases. When inspiratory effort decreases at endinspiration, inspiratory flow decreases and then stops, causing the integrator to reset and the assistance to terminate. Two levels of ventilatory support were used during PSV and PAV Pressure levels of 10 and 20 cmHs0 were applied during PSV, whereas 40 and 80% of elastic and resistive assistance were given during PAV To induce acute hypercapnia with a consequent variation in patient ventilatory requirement, a fixed dead space of 150 ml was applied between the pneumotachograph and the Y piece of the ventilator circuit (30). Measurements were obtained before (dead space off) and after (dead space on) application of dead space. During dead space on, an increase in PET,,~ of 30% of its baseline value was considered as the target value. The necessary time to obtain this result amounted to 8-10 min (9.11 t 0.11 min). Ventilatory modes, levels of assistance, and dead space were randomly applied. Measurements were obtained from 20-25 breaths in each experimental condition once a stable breathing pattern was observed. During dead space on, arterial blood samples and were obtained 15-20 min (18.11 t measurements of PETIT 0.18 min) after the target increase in PETIT had been obtained and a stable PETE-, level was observed at both levels of PSV and PAV The fractional inspired 02 concentration remained constant throughout the study period. Data analysis. Inspiratory time (TI), expiratory time (TE), and total breathing cycle time (TT) were determined from the V tracing. Tidal excursions of Pes (Apes) were also determined. During PSV and PAV, inspiratory pulmonary resistance (RL) and el as t ante (EL) were calculated by using the Mead and Wittenberger technique (21). Briefly, RL and EL were calculated by fitting the equation of motion of a singlecompartment model by using multilinear regression, as follows

pleural pressure required for initiation of inspiratory V is equal to the opposing level of elastic recoil pressure present at end expiration and/or the cessation of expiratory effort (13,22). Pressure-time product (PTP) per breath (PTP/b) for the inspiratory muscle was measured as the area under the Pes signal from the beginning of the inspiratory deflection to the end of inspiratory V relative to the Pes tracing obtained during the period of controlled mechanical ventilation and respiratory muscle relaxation and assuming that elastance of the chest wall (CW) was linear within the VT range (27). The static CW line was placed by assuming that the endexpiratory elastic recoil pressure of the CW was equivalent to the end expiratory position of Pes during tidal breathing (12). This assumption was verified by occluding the airway at end expiration during the four experimental conditions and demonstrating that the end-expiratory position of Pes under occluded breaths was essentially identical to that during unoccluded breaths. PTP per minute (PTP/min) was calculated as PTP/b multiplied by respiratory rate. PTP per liter of iTE (PTP/I) was calculated as PTP/min divided by i7E. The level of dyspnea during the application of dead space was evaluated at both levels of PSV and PAV Patients were instructed to indicate any change in the sense of breathlessness by pointing to a bidirectional ordinal scale (0 = no change; - l/+ 1 = very slight worsening/improvement; - 2/+ 2 = slight worsening/improvement; - 3/+ 3 = moderate worsening/im-4/+4 = marked worsening/improvement; provement; -5/+5 = very marked worsening/improvement) (4,23). StatisticaL anaZysis. Values are expressed as means t SE. Values obtained at different CO2 levels and with different modes of partial ventilation were compared by using the analysis of variance for repeated measures. Regression analysis was performed with the least squares method. A value of P < 0.05 was considered statistically significant. RESULTS

APL = RL.~&~~ + EL. AV

(1)

where APL is inspiratory change in transpulmonary pressure (calculated by subtracting Pes from Pao), AV is inspiratory change in lung volume between end-expiratory lung volume and VT, and Vpeak is peak inspiratory flow (14). The level of intrinsic PEEP during different levels of PSV and PAV (PEEPidyn) was measured as the negative deflection in Pes from the onset of inspiratory effort to the point of zero V (13, 22). This approach is based on the assumption that change in

Causes of ARF, sex, age, days of mechanical ventilation, and static respiratory mechanics during controlled mechanical ventilation obtained on the day of the study are shown in Table 1 for the individual patients. Arterial O2 (Pao,) and CO2 (Paco2>pressures and pH values just before the study are also indicated.

Table 1. Patient characteristics

Patient No. Sex Age, yr
Paoz*,

Cause

of ARF

FIO,

Torr

pace:!*, Torr

pH*

MIP, cmH20

Est,rsf, cmH20/l

RT,rst, cmH20 - 1-l - st

PEEPis, cmH20

7
Day

1 2 3 4 5 6 7 8 9 10 11 12

F F M F F M M M F F M M

37 62 41 28 21 51 67 49 57 58 25 38

Sepsis Pneumonia Sepsis Multiple Pancreatis, Multiple Peritonitis, Multiple Pneumonia Pneumonia Multiple Multiple

trauma laparotomy, trauma laparotomy, trauma

sepsis sepsis

trauma trauma

0.4 0.4 0.3 0.4 0.3 0.4 0.4 0.3 0.4 0.3 0.4 0.4

91 98 118 115 109 110 134 105 109 108 110 98

43 44 39 41 42 41 39 38 41 35 42 39

7.42 7.41 7.40 7.41 7.40 7.40 7.46 7.41 7.42 7.43 7.40 7.43

-30 -37 -41 -52 -41 -38 -37 -40 -42 -44 -31 -33

25 20 30 23 20 20 18 20 22 21 23 20

14 16 13 15 15 18 17 16 18 15 12 14

0.55 0.38 0.37 0.51 1.12 0.59 3.31 1.15 0.87 1.02 1.02 0.55

4 8 4 8 9 14 10 8 8 5 11 14

* Data obtained during pressure-support ventilation just before study; f data obtained during short trial of respiratory muscle controlled mechanical ventilation previous to experimental procedure. ARF, acute respiratory failure; FI+ fraction of inspired ventilation; Pao,, arterial 02 tension; Pace,, arterial CO2 tension; MIP, maximal inspiratory pressure; Est,rs, static elastance system; Rqrs, total resistance of respiratory system; PEEPi,t, static intrinsic positive end-expiratory pressure.

inactivity and 02 mechanical of respiratory

EFFECTS

OF CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

429

Table 2. Effects of dead space on Pace, and PETIT,! at different levels of PSV and PAV
Pace,, Torr PSV 20 cmHz0 Dead space Dead space PSV 10 cmHz0 Dead space Dead space PAV 80% Dead space Dead space PAV 40% dead space Dead space off on off on off on off on 42.58 58.22 + 2.47 + 2.46
PETIT,,

Torr 2.89 2.94*

Pace, - PETITE, Torr 5.57 + 0.53 9.215 1.01* 5.53 + 0.49 9.96 + 1.21* 3.72 t 0.88 5.20 + 0.89* 3.87 + 0.41 5.84 + 1.01*

37.015 49.01+ 46.46 61.36

51.99 + 2.54 71.32 k 2.44* 44.23 + 1.98 58.36 IL 2.25* 50.88 IL 2.87 69.02 5 2.10*

2 2.01 + 2.20*

40.512 2.72 53.16 + 3.01* 47.01+ 1.91 63.18 k 2.99*

Data are means + SE. PSV, pressure-support ventilation; PAV, proportional-assist ventilation; PETCO,, end-expiratory CO2 tension. *P < 0 .001 dead space off vs. dead space on, analysis of variance (ANOVA) for repeated measures.

The effects of application of dead space on Pace, and PETCO, values are shown in Table 2. As expected, application of dead space significantly increased both Pa co, and PETCO,values at both levels of ventilatory support with both PSV and PAV Pace, and PETCog values during PSV 20 cmH20 and 80% PAV and PSV 10 cmH20 and 40% PAV were not different during either dead space off or dead space on. Experimental records in a representative patient during high (Left) and low (right) levels of PSV (Fig. 1) and PAV (Fig. 2) are shown. As can be seen, at both levels of PSV, application of dead space induced a relevant increase in respiratory rate. Tidal swings in Pes, VT, and Vpeak increased, although slightly; Pao remained unchanged (Fig. 1). On the contrary, when

dead space was applied at both levels of PAV, tidal excursions in Pes, VT, Pappl, and V markedly increased, whereas respiratory rate remained unchanged (Fig. 2). When dead space was applied, VE increased (P < 0.001) during 20 and 10 cmH20 PSV (Table 3). Similarly, VE increased (P < 0.001) during 80 and 40% PAV (Table 4). However, the mechanisms leading to the increase in VE due to the acute rise in PETITE after the application of dead space differ between PSV and PAV In fact, during 20 and 10 cmH20 PSV, application of dead space increased VT by 13 t 1 and 21 t 2% during 20 and 10 cmHBO PSV, respectively, whereas respiratory rate increased 48 t 4 and 103 t 5% during 20 and 10 cmHBO PSV, respectively. A reduction in TI, TE, TT, and TI/TT was observed with the application of dead space during PSV, whereas Pappl remained unchanged and mean Vand Vpeakincreased at both levels of PSV (Table 3). When dead space was applied during PAV, VT remarkably increased 32 t 1 and 66 t 2% during 80 and 40% PAV, respectively, whereas respiratory rate, TI, TE, TT, and TI/TT remained unchanged. Pappl and mean Vand iTpeak increased after application of dead space at both levels of PAV (Table 4). Application of dead space did not affect EL and RL at low and high levels of ventilatory support during either PSV and PAV On the other hand, PEEPidyn increased significantly (P < 0.001) at both levels of PSV and PAV However, during PSV 10 and 20 cmHzO and hypercapnia, PEEPidyn was, respectively, 5 and 10 times larger than during normocapnia, whereas during 80 and 40% PAV, PEEPidyn was, respectively, 1 and 3 times larger during hypercapnia than during normocapnia (Tables 3 and 4).

A
dead space off Flow U-JS)
1

B
dead space on dead space off deadspaceon

0I

AV /T \

0.4

1
..I -,,,

lo01

Fig. 1. Experimental records illustrating effects of application of dead space at 20 (A) and 10 (B) cmH20 of pressure support in a representative patient. From top to bottom: flow, inspiratory change in lung volume (AV), airway opening pressure (Pao), and tidal excursions in esophageal pressure (Apes) signals are shown. PSV, pressure-support ventilation. Horizontal dashed lines indicate zero reference lines for flow and Pao.

430

EFFECTS

OF

CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

A
dead space Flow VJS) 1 0I

off

dead space on

dead space off

dead space on

Fig. 2. Experimental records illustrating effects of application of dead space at 80 (A) and 40% (B) cmH20 of proportional-assist ventilation (PAV) in same patient as in Fig. 1. Abbreviations and dashed lines are defined as in Fig. 1.

AV (L)

o-4

Pa0 @mH20)

lo oI

Apes
(&mH 2 0)

1o
I

The effects of application of dead space on tidal excursions in Pes (Apes), PTP/b, PTP/min, and PTP/l during PSV and PAV are shown in Tables 5 and 6, respectively. As can be seen, application of dead space significantly (P < 0.01) increased APes and PTP/b by Table 3. Effects of dead space on breathingpattern and mechanics at different PSV levels
PSV 20 cmHzO Dead space off
VT,

12 t 1 and 19 t 1% and by 13 5 1 and 20 t 2% during 20 and 10 cmHzO PSV, respectively (Table 5). A significant (P < 0.001) increase in APes and PTP/b after the application of dead space was also observed during 80% PAV (28 t 2 and 27 t 2%, respectively) and during 40% PAV (30 t 2 and 63 + 2% respectively) (Table 6). The increase in PTP/min after the application of dead space amounted to 79 ? 4 and 168 t 4% during 20 and 10 cmH20 PSV, respectively, and was significantly (P < Table 4. Effects of dead space on breathing pattern and mechanics at different PAV levels
PAV 80%, Dead Space Off
VT,

PSP IO cmH,O Dead space off Dead space on

Dead space on

liters

TI, s TE, s TT, s TI/TT

0.84 1.85 2.43 4.34 0.42

r 0.03 IT 0.03 t 0.03 + 0.03 + 0.01

0.95 1.07 1.91 2.97 0.36

2 0.03* -+ 0.02* t 0.02* t 0.02* + O.Ol*

0.57 1.33 2.35 3.64 0.35

+ + t + t

0.04 0.03 0.03 0.02 0.01

0.69 0.58 1.27 1.85 0.30

+ 5 + 5 +

0.03* 0.03* 0.02* O.Ol* O.Ol*

PAV 8O%, Dead Space On

PAV 40%, Dead Space Off

PAV 40%, Dead Space On

f, breaths/ min iTE, Vmin

VT/T& l//s

liters

qleak, l/s P max7 cmH20 21.23 + RL, cmHz0. l-is 15.112 EL, cmHzO/l 20.01? PEEPi dyn 7 cmHz0 0.65 +

13.66 + 1.61 11.49 t 0.46 0.46 + 0.02 1.18t0.03 0.11

20.20 + 1.62* 19.19 + 0.51 0.88 t 0.04* 1.37t0.02* 21.54 + 0.14

16.415 1.65 9.32 + 0.55 0.45 IL 0.03 0.68+0.04 10.212 0.13

33.17? 1.97* 22.88 + 0.40* 1.18 2 0.02* 1.38+0.03* 10.18 2 0.16

TI, s TE, s TT, s TI/TT

0.62 0.97 2.07 3.05 0.32

it k + + 2

0.19 0.11 0.19 0.12 0.09

0.82 + 0.18* 1.012 0.12 2.14 + 0.16 3.18? 0.11 0.312 0.11 18.84 + 2.78 15.50 + 1.19* 0.82 + 0.19* 1.07+0.11* 25.47 + 1.64*

0.41+ 0.17 0.96 + 0.14 2.05 + 0.13 2.98 + 0.11 0.31+ 0.09 20.112 3.13 8.25 +: 1.17 0.42 t 0.14 0.64t0.08 12.31 IT 3.01

0.68 1.02 2.04 3.03 0.33

2 + + + +

0.18* 0.11 0.19 0.10 0.12

f, breaths/ min i/TE, l/min

vT/TI, l/s vpeak, us P max 7

10.1 3.11 0.03

14.5821.11 21.03 4.212 + 2.05 0.51*

15.03rl.06 20.54 2 2.61

15.18r0.99 20.19? 2.09

19.60 + 2.84 12.24 2 1.06 0.64 2 0.15 0.96r0.10 16.23 It 1.48

l

19.79? 2.47 13.55 t 1.38* 0.65 + 0.15* 0.84t0.10* 20.21-+ 2.84*

0.53 + 0.05

4.85 IL 0.88*

TI, inspiratory time; TE, Data are means + SE. VT, tidal volume; expiratory time; TT, total breath cycle duration; TI/TT, ratio of TI to . flow; f, respiratory frequency; VT/TI, mean T T ; Vpeak, P eak inspiratory inspiratory flow; iTE, minute ventilation; Pm,,, maximal airway pressure; RL, pulmonary resistance; EL, pulmonary elastance; intrinsic positive end-expiratory pressure. *P < PEE Pi dyn 7 dynamic 0.001 dead space off vs. dead space on, ANOVA for repeated measures.

cmH20 RL, cmH20 l-l*s EL, cmHzO/l

PEEPidyn,

15.05 + 1.06 19.85 + 2.67 0.68 + 0.02

14.97 + 1.15 20.51+ 2.10

15.63 + 1.33 19.88 t 2.03 0.58 k 0.04

15.24 + 1.29 20.11* 2.99

cmH20 Data ANOVA

1.55 +_ 0.09*

2.89 + 0.03*

are means + SE. *P < 0.001 for repeated measures.

dead space off vs. dead space on,

EFFECTS

OF CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

431

Table 5. Effects of dead space on inspiratory effort at different PSV levels

PSV 20 cmH20 Dead space off Dead space on PSV Dead space off

muscles
10 cmH20 Dead space on

Apes, cmHz0 PTP/b, cmHz0.s PI?Plmin, cmHz0. smin+ PTPI, cmHzO* s-i

3.69 + 0.12 3.32 k 0.09

4.15 + O.ll* 3.98 + O.ll*

9.43 5 0.23 6.7850.12

10.61 k 1.05* 8.09 + 0.19*

44.46 t 1.21

80.36 + 1.12*

110.11 + 3.77

294.85 2 4.04*

3.87kO.16

4.18kO.08

11.58 + 0.12

13.01+

0.12*

Data are mea ns t SE. Apes, tidal excursions in esophageal PTP/min, prespressure; PTP/b, pressure-time product per breath; sure-time product per minute; PTP/I, pressure-time product per liter of VE. *P < 0.001 dead space off vs. dead space on, ANOVA for repeated measures.

0.001) larger than that observed during 80 and 40% PAV (22 -+ 2 and 63 IT 3%, respectively). Application of dead space induced a significant (P < 0.001) increase in PTP/l, amounting to 8 t 1 and 12 ? 1% during 20 and 10 cmHBO PSV, respectively (Table 5>, whereas it had no effect on PTP/l at both levels of PAV (Table 6). When asked to indicate changes in the degree of breathlessness during the application of dead space with respect to the preceding dead space off condition, all patients reported an increase in dyspnea. The mean values of the group were -3 05 + 0.21 and -4.85 t 0.35 during 20 and 10 cmH,OPSV, respectively. These values were significantly lower than those reported during 80% PAV (- 1.33 -+ 0.25, P < 0.01) and 40% PAV (-2.18 t 0.31, P < 0.01).
DISCUSSION

Our data show that, in mechanically ventilated patients during the weaning phase, the strategy used to compensate for increased ventilatory requirements due to CO2 stimulation differs substantially between PSV and PAV During PSV, despite an increase in inspiraTable 6. Effects of dead space on inspiratory effort at different PAV levels
PAV 80%, Dead Space Off PAV 80%, Dead Space On PAV 40%, Dead Space Off

muscle
PAV 40%, Dead Space On

Apes, cmHz0 PTP/b, cmH2Os PTP/min, cmHz0 wmin-i PTPA, cmHz0 SW Data ANOVA

3.16 + 0.49 2.24 k 0.10

l

4.06 + 0.51* 2.85 Ifi O.ll*

9.02 + 0.65 4.76 + 0.09

11.78 + 0.32* 7.76 k 0.12*

43.94
l

+ 1.06

53.60

+ 1.27*

95.30

+ 1.70

153.55

k 2.77*

3.56 -+ 0.06

3.60 2 0.07 dead

11.55 + 0.08

11.69 + 1.21

are means t SE. *P < 0.001 for repeated measures.

space off vs. dead space on,

tory muscle effort, the increase in iTE due to acute hypercapnia is mostly obtained through an increase in respiratory rate. On the contrary, during PAV, the increase in VE is essentially obtained through an increase in VT due to the larger inspiratory muscle effort, leaving respiratory rate unchanged. When the energy cost (evaluated as PTP/min and PTP/l) of such different strategies is compared, our data show that a larger amount of inspiratory muscle effort is required to increase VE in response to the hypercapnic stimulation during PSV than during PAV Furthermore, patient discomfort during CO2 stimulation seems lower during PAV than during PSV. Before the results of the present study can be discussed, some general considerations concerning the experimental design are required. Both PSV and PAV are capable of unloading the respiratory muscles (6, 7, 17, 32, 35). In fact, with either mode, the amount of work of breathing that is taken by the ventilator can be varied from slight to nearly total. In the present investigation, there was no attempt to make high and low levels of ventilatory assistance during PSV and PAV comparable concerning levels of Pappl, peak flow, TI and TE durations, and respiratory rate. Therefore, we are not allowed to compare the performance of PSV and PAV in reducing inspiratory muscle effort. On the other hand, the hypercapnic stimulation of the respiratory drive obtained by application of dead space was similar during 20 cmH20 PSV and 80% PAV and during 10 cmHBO PSV and 40% PAV As a consequence, we may first assess the different mechanisms leading to the increase in irE and, second, compare patient workload necessary to compensate for increase ventilatory requirements during PSV and PAV General consensus holds that during mechanical ventilation, the ventilator should cooperate with patients and take on exactly the amount of work the patients cannot do by themselves. That is why supported ventilation is now being favored rather than controlled ventilation (29). PSV (6, 7, 17) and PAV (32, 35) are two recently introduced methods for partial ventilatory support. They are intended for patients whose respiratory drive is normal or high but who have difficulty in sustaining an adequate level of ventilation on their own and in whom an abnormal relationship between effort and ventilatory consequences (neuroventilatory coupling) (37) is present. The basic feature in common between PSV and PAV, which distinguishes them from other methods of partial ventilatory support, is that in both modes the target is to assist each inspiration; each inspiratory effort that succeeds in decreasing airway pressure or flow below the triggering value causes the ventilator to increase airway pressure. The fundamental difference between PSV and PAV is the algorithm that governs Pappl behavior once the ventilator is triggered. During PSV, Pappl follows a nearly square pattern. Differences in level of effort beyond the inspiratory triggering threshold should not influence the level of Pappl, which should remain constant throughout inspiration until a cycling-off condition is met. The latter is most commonly a de-

432

EFFECTS

OF CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

crease in flow rate below a threshold value (Vthr,) (6, 7, 17). With PAV, Pappl changes as a function of patient effort; the more the patient pulls, the more the ventilator delivers flow and volume. This relationship applies as the level of effort changes from breath to breath. In addition, the pattern of Pappl reflects the pattern of pressure generated by the patients respiratory muscle within each inspiration. Neither the level of pressure nor flow or volume is set. Rather, what is set is the proportion between instantaneous Pmus and instantaneous Pappl(32,35). In our study, an increase in ventilatory requirement was induced by acute hypercapnia consequent to the application of a fixed dead space (30). An increase in VE is the physiological response to hypercapnia (9). A number of workers have shownthat, in normal subjects, this physiological increase in VE is almost entirely the result of an increase in VT, whereas the increase in respiratory rate seems to be less relevant (1,5,&g). In a more recent study, Yan et al. (31) confirmed these findings and showed that the increase in VT consequent to acute hypercapnia is due to an increase in amplitude of esophageal and transdiaphragmatic tidal swings. Our data show that a physiological response to CO2 stimulation similar to that described in normal subjects (1,5,8,9) was observed only during PAV Application of dead space significantly (P < 0.001) increased APes and PTP/b by 28 t 2 and 27 2 2%, respectively, during 80% PAV and by 30 + 2 and 63 t 2% during 40% PAV (Table 6). This increase in inspiratory muscle effort produced a concomitant rise in VT (Table 4). As a consequence, during PAV, the physiological neuroventilatory coupling (i.e., the patients capability to integrate respiratory drive, inspiratory muscle effort, and VT according to the different ventilatory requirements)

was preserved (37). In a computer simulation based on a model described previously that aims to estimate the ventilatory consequences of different patterns of respiratory motor output in spontaneously breathing subjects under different physiological and pathological conditions (34, 36), Younes (33) studied the effect of progressively increasing peak inspiratory effort on the time course of VT, flow, and Pappl during PAV He showed that the patients ability to modify VT, V, and Pappl through changes in his own effort (VT vs. Pmus slope: neuroventilatory coupling) progressively improved as the proportion of the assistance (Pappl vs. Pmus relationship slope) was increased. To verify Youness computer simulation in the clinical setting, we plotted Pmus (evaluated as PTP/b) against VT, peak flow, and Pappl (evaluated as P,,,) for 15-20 consecutive breaths during dead space off and dead space on conditions and at both levels of PAV in each patient (Fig. 3). Confirming Youness simulations, we found a significant (P < 0.0001) correlation between inspiratory muscle effort and VT, peak flow, and Pappl. As a consequence of the increase in assistance from PAV 40 to SO%, the slopes Of VT vs. PTP/b, peak flow vs. PTP/b, and Pappl vs. PTP/b relationships increased, indicating that neuroventilatory coupling improved as the PAV level rose. These data show that, with PAV, the patients ability to modulate ventilatory output through variations in inspiratory effort is preserved and actually enhanced, the ventilator essentially becoming an extension of the patients own respiratory muscles. During PSV, the significant (P < 0.001) increase in inspiratory muscle effort consequent to acute CO2 stimulation of the respiratory drive (Table 6) was followed by a slight increase in VT (13 t 1 and 21 t 2% during 20 and 10 cmH20 PSV, respectively), and the

A
1.2 1 1.0 1 1.2, I
301

%
g 0 20-

GJ 0.8 ? <1 0.6 -

R=0.99 slope = 0.34

0.6 1

R = 0.97 slope = 0.18

R = 0.99 slope = 4.3 1

1.2 z 1.0

R = 0.95 slope = 0.17

0 4 8 12

0.4

R = 0.92 slope = 0.085

4 8 12 16 20 0 4 8

R = 0.99 slope = 2.31

12 16 20

16 . set)

20

PTP/b Fig. 3. Effects of PAV PTP/b, of increasing pressure-time inspiratory product

(cmHzO

PTP/b

(cmHzO

. set) 80

PTP/b

(cmH20

. set)

effort on AV, flow, and maximal airway pressure (P,,,) during per breath. See text for further details. 0, Dead space off;

(A)

and

40% (B)

l , dead space on.

EFFECTS

OF CHANGES

IN RESPIRATORY
tiOll

DRIVE:

PSV VS. PAV

433

rise in VE was essentially based on the increase in respiratory rate (48 ? 4 and 103 t 5% during 20 and 10 cmHzO PSV, respectively), which caused a reduction in TI, TE, and TI/TT (Table 3). Recently, Scheid et al. (28) tested CO2 sensitivity in eight healthy volunteers during PSV administrated by mouthpiece. Scheid et al. found that, at 5 cmHgO PSV, VT increased linearly with the slope of this relationship being nearly the PETCO,~ same as the one obtained during spontaneous breathing. At 10 cmH20 PSV, VT appeared to be independent of PETIT until a threshold of 38 Torr was exceeded. In sharp contrast with VT, respiratory rate did not reveal a consistent dependence on CO2 levels (28). Direct measurements of inspiratory muscle effort were not obtained in the study of Scheid et al. (28). However, the authors (28) concluded that, as previously shown in spontaneously breathing subjects (31), COz stimulation increased VT during PSV due to greater inspiratory muscle effort. To explain differences between Scheid et al. (28) and our results, the effect of changes in peak inspiratory effort on VT and TI during PSV should be considered. With PSV, the effects of a given effort on VT and TI depend critically on the VT and TI that would occur at the same level of PSV in the absence of inspiratory muscle effort (18). Under these circumstances, the relationship between Pappl and generated flow and volume is described by the following equation

ill TImech with increase in Vpeak and no change in VT. On the other hand, when inspiratory muscle effort is sufficiently intense and TI,e,, equals TImech, VT will become sensitive to patient effort (33). In our patients, TI mech(calculated according to Eq. 4) amounted to 2.02 t 0.01 and 1.50 t 0.02 s during 20 and 10 cmHzO PSV, respectively. Our data show that TI,e,, (obtained from the flow signal) (3) was shorter (P < 0.01) than TI mech. Therefore, confirming Youness (33) simulation, in our patients an increase in inspiratory muscle effort did not produce a large increase in VT at either level of PSV. On the other hand, Scheid et al. (19) found that, when the PSV setting was higher relative to the impedance of the relaxed respiratory system, CO2 had no effect until it reached a level that increased respiratory drive. In our patients, respiratory impedance was higher than in the normal subjects studied by Scheid et al. (19). Therefore, the level of machine support was insufficient to uncouple the duration of inspiratory effort from the duration of the machine inflation cycle. Some recent data from Puddy and Younes (25) suggest that inspiratory V has a positive chronotropic effect and that it may also augment inspiratory drive in normal subjects ventilated through a mouthpiece (25). In our patients, we found that, during dead space breathing, mean V and Vpeakwere significantly (P < 0.01) higher during PSV (0.88 ? 0.02 and 1.37 t 0.02 l/s during 20 cmHBO PSV and 1.18 t 0.02 and 1.38 t 0.03 Pappl = Pres + Pel (2) l/s during 10 cmHzO PSV, respectively) than during PAV (0 82 + 0 .19 and 1.07 ? 0.11 l/s and 0.65 t 0.15 where Pres and Pel represent the amount of Pappl and 0.84 ? 0.10 l/s, respectively). These data may also necessary to overcome patient RT,rs and Est,rs, respec- contribute to explain the predominant rate response tively (18). If resistance and elastance are constant, Eq. observed during PSV and hypercapnic chemoreceptor 2 becomes stimulation. However, whereas in Puddy and Youness study Vpeak was preset (25), during PSV V is deterPappl = VaR~,rs + AVEst,rs (3) mined by the patients respiratory mechanics and inspiDuring PSV, taking the case of a patient who makes ratory muscle effort and by the pressure level set on the just enough effort to trigger the ventilator, after an ventilator (Eq. 3). Other mechanisms, apart from posiinitial peak, flow will start to decline and volume to rise tive chronotropic effect of Vpeak,must be therefore taken according to an exponential function (18). Under these into account to explain the rise in respiratory frequency circumstances, TI is determined exclusively by the time observed during PSV and hypercapnia. With added dead space, PEEPidyn was higher (P < taken for the exponentially declining flow to reach Vthre, 0.001) during PSV than during PAV (4.21 t 0.51 vs. at which the cycling between inspiration and expiration occurs [mechanical TI (TI~~~~)] (18, 33). T1mech can 1.55 t 0.09 cmHsO during 20 PSV and 80% PAV, therefore be calculated according to the following equa- respectively, and 4.85 ? 0.88 vs. 2.89 t 0.03 cmHBO during 10 cmHzO PSV and 40% PAV, respectively). This tion (18,33) could have also influenced the response to hypercapnia TI mech = 7 In (peak flOWtheor/i/Tthre) (4) observed during PSV with respect to the VT position on the volume-pressure curve (particularly considering where 7 is the time constant of the progressive decline that PEEP,,, is likely to underestimate the increase in in V during inspiration, calculated as RT,rs/Est,rs; and lung volume) and to the high values of EL observed in is the theoretical value of initial flow our patients. Peak flOWtheor obtained by dividing the experimental values of conIn our study, to evaluate the efficacy of the neurovenstant Pappl by the experimental value of RT,rs, suppos- tilatory coupling during PSV, the VT vs. PTP/b, Vpeakvs. ing that the respiratory system behaves as a monocom- PTP/b, and Pappl vs. PTP/b relationships for each partment linear model (18). In a computer simulation, individual patient were calculated at both levels of PSV Younes (33) studied the effect of progressively increas- by plotting the values obtained from 15-20 consecutive ing peak inspiratory effort on the time course of flow breaths during dead space off and dead space on and volume during PSV. He showed that, when the conditions (Fig. 4). Changes in inspiratory muscle effort patient TI [neural TI (TI,e,,)] is less than TImech, progres- were not correlated with VT Pappl, whereas a signifisive increases in peak inspiratory effort cause a reduc- cant (P < 0.0001) positive correlation was found bel

434

EFFECTS

OF CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

3 ; u

l.O0.80.6 0.4R = 0.73 slope = 0.051 16

z w 3 0 E

1.0 0.8 0.6 R = 0.86 slope = 0.066 16 R = 0.51 slope, = 0.18 16 20

B
s > Q

1.21.0 2 w

1.2 1.0 -

3o

Gc f 20
0 P I 4 I 8 (cmH20 R = 0.96 slope = 0.076 1 12 I 16 . set) I 20

0.80.60.4 0.2 i 0 I 4 I 8 (cmH,O R = 0.95 slope = 0.032 , 12 I 16 . set) 1 20

3 0.8 Fr, 0.6 0.4 1 0

3 10 &E 1
0 i

1
0

R = 0.35 slope = -0.064

I I 1 a I

8 (cmH,O

12

16 . set)

20

PTP/b Fig. 4. Effects Abbreviations

PTP/b

PTP/b

of increasing inspiratory effort on AV, flow, and P,, are defined as in Fig. 4. See text for further details.

during

20

(A) and 10 (B) cmH20

of PSV.

tween PTP/b and peak flow; a slightly significant correlation (P < 0.05) was found between VT and PTP/l. It follows that, with PSV, the patients ability to modulate VT according to ventilatory requirements by varying inspiratory muscle effort is limited, whereas a discrepancy between the respiratory rates of the patient and of the ventilator probably occurs. Besides, duration and Vpeak seem more sensitive than VT as indicators of the changes in intensity of patient effort. McGregor and Becklake (20) have shown that the Oz cost of breathing is closely related to the mean pressure developed by the respiratory muscles. The latter, i.e., the total energy cost of breathing, was evaluated in the present study as the mean pressure developed by the respiratory muscles per minute (PTP/min) and per liter of VE (PTP/l) (12). In our patients, CO2 stimulation increased PTP/min and PTP/l by 79 ? 4 and 8 t l%, respectively, during 20 cmH20 PSV and 168 2 4 and 12 t I%, respectively, during 10 cmH20 PSV These changes were significantly higher than those observed after the application of dead space during 80% PAV (PTP/min and PTP/l increased to 22 t 2 and 1 t l%, respectively; P < 0.01 vs. 20 cmHBO PSV) and 40% PAV (PTP/min and PTP/l increased to 60 t 2 and 1 t l%, respectively; P < 0.001 vs. 10 cmHzO PSV). In addition, during acute hypercapnia, the absolute values of PTP/ min and PTP/l were significantly higher (P < 0.01) at both levels of PSV than those observed at the corresponding levels of PAV (Tables 5 and 6). The reduced sense of breathlessness during acute hypercapnia observed at both levels of PAV is therefore consistent with the lower inspiratory muscle effort required to increase VE (15). Respiratory rate is in fact normally set at a particular value, which is less costly in terms of O2 cost of breathing (24). Under these circumstances, the increase in respiratory rate necessary to increase VE

after CO2 stimulation observed during PSV did not provide optimal energy conservation and was therefore accompanied by increased work of breathing and greatest patient discomfort. Application of dead space significantly (P < 0.001) increased PEEPidyn at both levels of PSV and PAV (Tables 3 and 4). PEEPid, is a common finding in mechanically ventilated patients, but different pathophysiological mechanisms may be responsible (26). In fact, in patients with chronic obstructive pulmonary disease (COPD), PEEP,,, may be caused by dynamic hyperinflation and expiratory flow limitation. Alternatively, under conditions of very high VE PEEPidyn may be caused because the lungs do not have time to reach their equilibrium point. Besides, when expiratory muscle recruitment occurs at end exhalation, flow persists to the end of the expiratory cycle, and an end-expiratory gradient of alveolar to central airway pressure is present without PEEPidyn and an increase in end-expiratory lung volume (26). Patients included in the present investigation did not have a previous history of COPD. Therefore, the observed increase in with hypercapnia may be attributed to both PEEPidyn the increase in i7E (26) and/or to the expiratory recruitment of the abdominal muscles (31). However, the effects of both ventilatory support on PEEPidyn are difficult to compare because the contribution of expiratory muscle activity may vary at the different levels of muscle unloading and with the different ventilatory modes (16). Despite the fact that baseline VT values and hypercapnit chemoreceptor stimulation were similar during both levels of PSV and PAV, ventilatory requirement during dead space breathing was much higher during PSV than during PAV (iTE amounting to 19.19 t 0.51 and 15.50 + 1.19 Vmin during 20 cmHzO PSV and 80%

EFFECTS

OF CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

435

PAV and to 22.88 t 0.40 and 13.55 of: 1.38 l/min during 10 cmHzO PSV and 40% PAV, respectively). With the assumption that our measurements were really taken during steady-state conditions, differences in dead space-to-VT ratio and/or CO2 production may explain such large differences in the ratio of VE to CO2 level observed between PSV and PAV (11). Confirming this hypothesis, we found that the increase in Pace, difference consequent to application of dead PER0 space2was significantly (P < 0.01) larger during PSV + 2% during 20 and 10 cmHzO PSV, (65 t 1 and 80 respectively) than during PAV (40 t 1 and 50 t 2 during 80 and 40% PAV, respectively). In conclusion, our data show that, in mechanically ventilated patients in whom ventilatory requirements are acutely increased by provoking hypercapnia through the application of dead space, the physiological capability to increase VE through changes in VT modulated by variations in inspiratory muscle effort was preserved only during PAV During PSV, the increase in VE due to the higher ventilatory requirements was obtained through an increase in respiratory rate, whereas VT and inspiratory muscle effort remained substantially unchanged. The compensatory strategy used to increase VE during PSV required greater muscle effort and caused more pronounced patient discomfort than during PAV It has still to be proved whether the physiological differences between PSV and PAV during the weaning phase have significant clinical implications.
The authors thank the physicians and nursing staff of the Policlinic0 Hospital for their valuable cooperation and Nellcor Puritan Bennett for its support. They also thank Mary V. C. Pragnell for help in revising the manuscript. This work was supported in part by Consiglio Nazionale delle Ricerche Grant 94.02325.CT04. Address for reprint requests: V. M. Ranieri, Istituto di Anestesiologia e Rianimazione, Policlinico, Piazza G. Cesare, 70 124 Bari, Italy (E-mail: rianima@cimedoc.uniba.it). Received 21 September 1995; accepted in final form 9 February 1996.

REFERENCES 1. Askanazi, J., J. Milic-Emili, J. R. Broell, A. I. Hyman, and J. M. Kinney. Influence of exercise and CO2 on breathing pattern of normal man. J. Appl. Physiol. 47: 192-196,1979. 2. Baydur, A., P. K. Behrakis, W. A. Zin, M. I. Jaeger, and J. Milic-Emili. A simple method for assessing the validity of the esophageal balloon technique. Am. Rev. Respir. Dis. 126: 788791,1982. 3. Bellmare, IF., and A. Grassino. Effect of pressure and timing of contraction on human diaphragm fatigue. J. Appl. Physiol. 53: 1190-11951982. 4. Borg, G. A. K Psychopathological basis of perceived exertion. Med. Sci. Sports Exercise 14: 377-381, 1982. 5. Bradley, G. W., C. von Euler, I. Martilla, and B. P. Roos. Steady state effects of CO2 and temperature on the relationship between lung volume and inspiration duration (Hering Breuer threshold curve). Acta PhysioZ. Stand. 92: 351-363,1974. 6. Brochard, L. Pressure support ventilation In: Principles and Practice of Mechanical VentiZation, edited by M. J. Tobin. New York: McGraw-Hill, 1994, p. 239-257. 7.. Brochard, L., A. Harf, H. Lorino, and F. Lemaire. Inspiratory pressure support prevents diaphragmatic fatigue during weaning from mechanical ventilation. Am. Rev. Respir. Dis. 139: 513-521,1989. 8. Clark, F. J., and C. von Euler. On the regulation of rate and depth of breathing. J. Physiol. Lond. 222: 267-295,1972.

9. Cunningham, D. J. C., P. A. Robbins, and C. B. Wolff. Integration of respiratory responses to changes in alveolar partial pressures of CO2 and 02 and in arterial pH. In: Handbook of PhysioZogy. The Respiratory System. Control of Breathing. Bethesda, MD: Am. Physiol. Sot., 1986, sect. 3, vol. II, pt. 2, chapt. 15, p. 475-528. 10. DAngelo, E., E. Calderini, G. Torri, F. M. Robatto, D. Bono, and J. Milic-Emili. Respiratory mechanics in anesthetized paralyzed humans: effects of flow, volume, and time. J. AppZ. PhysioZ. 67: 2556-2564, 1989. 11. DuBois, A. B., A. G. Britt, and W. 0. Fenn. CO2 during the respiratory cycle. J. AppZ. Physiol. 4: 535-548,195l. 12. Fleury, B., D. Murciano, C. Talamo, M. Aubier, R. Pariente, and J. Milic-Emili. Work of breathing in patients with obstructive pulmonary disease in acute respiratory failure. Am. Rev. Respir. Dis. 131: 822-827,1985. 13. Haluszka, J., D. A. Chartrand, A. E. Grassino, and J. Milic-Emili. Intrinsic PEEP and arterial CO2 in stable patients with chronic obstructive pulmonary disease. Am. Rev. Respir. Dis. 141: 1194-1197,199O. 14. Kano, S., C. J. Lanteri, P. J. Pemberton, P. N. Lesouef, and P. D. Sly. Fast versus slow ventilation for neonates. Am. Rev. Respir. Dis. 148: 578-584, 1993. 15. Killian, K. J., and N. L. Jones. Respiratory muscles and dyspnea. CZin. Chest Med. 9: 237-248, 1988. 16. Lessard, M. R., F. Lofaso, and L. Brochard. Expiratory muscle activity increases intrinsic positive end-expiratory pressure independently of dynamic hyperinflation in mechanically ventilated patients. Am. J. Respir. Crit. Care Med. 151: 562-569, 1995. 17. MacIntyre, N. R., and N. E. Leatherman. Ventilatory muscle loads and the frequency-tidal volume pattern during inspiratory pressure-assisted (pressure-supported) ventilation. Am. Rev. Respir. Dis. 141: 327-331,199O. 18. Marini, J. J., P. S. Crooke, and J. D. Truwit. Determinant and limits of pressure-preset ventilation: a mathematical model of pressure control. J. AppZ. Physiol. 67: 1081-1092,1989. 19. Marini, J. J., T. C. Smith, and V. J. Lamb. External work output force generation during synchronized intermittent mechanical ventilation. Am. Rev. Respir. Dis. 138: 1169-1179,1988. 20. McGregor, M., and M. Becklake. The relationship of oxygen cost of breathing to respiratory mechanical work and respiratory force. J. CZin. Invest. 40: 971-980, 1961. 21. Mead, J., and J. L. Whittenberger. Physical properties of human lungs measured during spontaneous respiration. J. AppZ. PhysioZ. 5: 770-779, 1953. 22. Ninane, V., F. Rypens, J.-C. Yernault, and A. De Troyer. Abdominal muscle use during breathing in patients with chronic obstructive airflow obstruction. Am. Rev. Respir. Dis. 146: 16-21, 1992. 23. ODonnell, D. E., and K. A. Webb. Exertional breathlessness in patients with chronic airflow limitation. Am. Rev. Respir. Dis. 148: 1351-1357,1993. 24. Otis, A. B. The work of breathing. Physiol. Rev. 34: 449-458, 1954. 25. Puddy, A., and M. Younes. Effect of inspiratory flow rate on respiratory output in normal subjects. Am. Rev. Respir. Dis. 146: 787-789,199O. 26. Ranieri, V. M., R. Giuliani, G. Cinnella, C. Pesce, N. Brienza, E. L. Ippolito, V. Porno, S. B. Gottfried, and A. Brienza. Physiologic effects of positive end-expiratory pressure in COPD patients during acute ventilatory failure and controlled mechanical ventilation. Am. Rev. Respir. Dis. 147: 5-13, 1993. 27. Sassoon, C. S. H., R. W. Light, R. Lodia, G. C. Sieck, and C. K. Mahutte. Pressure-time product during continues positive airway pressure, pressure support ventilation, and T-piece during weaning from mechanical ventilation. Am. Rev. Respir. Dis. 143: 469-475,199l. 28. Scheid, P., F. Lofaso, D. Isabey, and A. Harf. Respiratory response to inhaled CO2 during positive inspiratory pressure in humans. J. AppZ. PhysioZ. 77: 876-882, 1994. 29. Slutsky, A. S., and the American College of Chest Physicians Consensus Conference. Mechanical ventilation. Chest 104: 1833-1859,1993.

436

EFFECTS

OF

CHANGES

IN RESPIRATORY

DRIVE:

PSV VS. PAV

30. Sly, P. D., J. H. T. Bates, T. Kochi, S. Okubo, and J. Milic-Emili. Frequency-dependent effects of hypercapnia on respiratory mechanics of cats. J. Appl. Physiol. 62: 444-450, 1987. 31. Yan, S., I. Lichras, S. Zakynthinos, and P. T. Macklem. Effect of diaphragmatic fatigue on control of respiratory muscles and ventilation during CO2 rebreathing. J. Appl. Physiol. 75: 1364-1370,1993. 32. Younes, M. Proportional assist ventilation, a new approach to ventilatory support. Am. Reu. Respir. Dis. 145: 114-120,1992. 33. Younes, M. Proportional assist ventilation, and pressure support ventilation: similarities and differences. In: Ventilatory Failure, edited by J. J. Marini and C. Roussos. Berlin: SpringerVerlag, 1991, p. 361-380.

34. Younes, M. Load responses, dyspnea and respiratory failure. Chest 97: 59S-68S, 1990. 35. Younes, M., A. Puddy, D. Roberts, R. B. Light, A. Quesada, K. Taylor, L. Oppenheimer, and H. Cramp. Proportional assist ventilation. Results of an initial clinical trial. Am. Rev. Respir. Dis. 145: 121-129,1992. 36. Younes, M., and W. Riddle. Relation between respiratory neural output and tidal volume. J. AppZ. Physiol. 56: 1110-1119, 1984. 37. Zechman, F. W., and R. L. Wiley. Afferent inputs to breathing: respiratory sensation. In: Handbook of PhysioZogy. The Respiratory System. Control of Breathing. Bethesda, MD: Am. Physiol. Sot., 1986, sect. 3, vol. II, pt. 1, chapt. 14, p. 449-474.