Evaluation of Respiratory Inductive Plethysmography in Controlled Ventilation

Evaluation of Respiratory Inductive Plethysmography in Controlled Ventilation* Measurement of Tidal Volume and PEEP-Induced Changes of End-Expiratory Lung Volume Peter Neumann, MD; Jorg Zinserling, MSc; Christian Haase; Michael Sydow, MD, PhD; and Hilmar Burchardi, MD, PhD

Study objective: To detennine the accuracy of respiratory inductive plethysmography (RIP) with a respiratory monitor (Respitrace Plus; NIMS Inc., Miami) operating in the DC-mode for the measurement of tidal volumes (VT) and positive end-expiratory pressure (PEEP)-induced changes of end-expiratory lung volume (aEELV) in patients with nonnal pulmonary function, acute lung injury (ALI), and COPD during volume-controlled ventilation. Design: Prospective comparison of RIP with pneumotachography (PT) for assessment of VT and with multibreath nitrogen washout procedure (N2 WO) for determination of aEELVas reference methods. Setting: Mixed ICU at a university hospital. Patients: Thirty-one sedated and paralyzed patients: 12 patients with nonnal pulmonary function mechanically ventilated after major surgery, 10 patients with respiratory failure due to ALI, and 9 patients with a known history of COPD ventilated after surgery or because of respiratory failure due to bronchopulmonary infection. lnteroentions: Stepwise increase of PEEP from 0 to 5 to 10 em H 2 0 and reduction to 0 em H 2 0 again. On each PEEP level, N 2 WO was perfonned. Measurements and main results: The baseline drift of RIP averaged 25.4±29.1 mUmin but changed over a wide range even in single patient measurements. Detennination of VT for single minutes revealed that 66.5% and 90.0% of all values were accurate within a range of ±10% and ±20%, respectively. The deviation for VT measurements between RIP and PT in patients with COPD was significantly (p<0.05) higher compared to patients with ALI or nonnal pulmonary function. The difference of aEELV between RIP and N 2 WO was 11.6±174.1 mL with correlation coefficients of 0.77 (postoperative and COPD patients) and 0.86 (ALI patients). However, just 25.8% and 46.2% were precise within ± 10% and ±20%, respectively. aEELV determination in COPD patients differed more between RIP and N2 WO than in the other groups, but this was not significant. Conclusion: In a mixed group of patients undergoing controlled ventilation, RIP using the Respitrace Plus monitor was not consistently precise enough for quantitative evaluation of VT and EELV when compared to our reference methods. This was most evident in patients with COPD. For long-tenn volume measurements, a better control of the baseline drift of RIP should be achieved. (CHEST 1998; 113:443-51) Key words: acute lung injury; COPD; end-expiratmy lung volume; respiratory inductive plethysmography; tidal volume Abbreviations: ALI=acute lung injury; CPAP=continuous positive ai1way pressure; Cst = quasi static compliance; EELV=end-expiratory lung volume; ~E ELV=change of end-expiratory lung volume; ~EELVN 2 wo=change of endexpiratory lung volume measured with nitrogen washout; ~EELVRIP =change of end-expiratory lung volume measured with RIP; Fio2 = fraction of inspired oxygen; FRC = functional residual capacity; N 2WO= multibreath nitrogen washout maneuver; Pe nd-e,l'iratory = end-expiratmy airway pressure; P peak =peak airway pressure; Pplateau=plateau ainvay pressure; PEEP =positive end-expiratory pressure; POSTOP=group of patients with normal pulmona1y function ventilated postoperatively after major surgery; PT=pneumotachography; Raw=ainvay resistance; RIP= respiratory inductive plethysmography; VT = tidal volume

R

espiratory inductive plethysmography (RIP) is a noninvasive method for determination of changes in thoracic volume. It is used to monitor

*From the Department of Anesthesiology, Eme rgency, and Intensive Care Medicine, University of Gi:ittingen, Germany. Manuscript received August 6, 1996; revision accepted June 2, 1997.

tidal volume (VT) and to detect changes in endexpiratory lung volume (dEELV). 1 End-expiratory lung volume (EELV) is decreased during acute lung injury (ALI ). 2 Ventilator settings aim to increase EELV by application of external positive end-expiratory pressure (PEEP)3 and intrinsic PEEP using inverse ratio ventilation. 4 In contrast, patients with COPD often have an increased functional residual CHEST I 113 I 2 I FEBRUARY, 1998

443

capacity. 5 During mechanical ventilation , it is considerably important to minimize hyperinflation of the lungs. Therefore, RIP might be a valuable tool to optimize respirator settings in mechanical ventilation by monitoring the EELV. Thus, RIP offers an attractive, noninvasive way of monitoring critically ill patients. However, its accuracy has been questioned.6 We compared RIP \vith pneumotachographic (PT) measurement of VT and with determination of 6.EELV by a multibreath nitrogen washout procedure (N2vVO ) in volume-controlled mechanically ventilated patients with ALI, COPD, and in postoperative patients (POSTOP group) who had no history or clinical signs of lung disease.

MATERIALS AND METHODS After approval of the protocol by th e ethical committee of the local m edical facu lty, 31 patients were included into the study. Patients undergoing elective surgery gave informed written consent to participate prior to the study; for all other patients, permission was obtained from the next of kin. Exclusion criteria were age younger than 18 years, fraction of inspired oxygen (Fio 2 ) > 0.6 to obtain an oxygen saturation of 90%, bronchopulmonary leakage, or circulatory instabili ty that would not allow a rapid change of the PEEP level. Measurements were performed in the a nesthesiologic ICU. All patients were deeply sedated, paralyzed, and mechanically ventilated. Volume-controlled ventilation was performed with constant flow, inspiratory:expiratory time ratio (I: E) of 1:2, respiratory rate of lO breaths/min, and VT adjusted to achieve normocapnia (PaC0 2 , 35 to 45 mm Hg). Twelve patients (POSTOP group) without history or evidence of lung disease were studied within 2 to 4 hafter major surgery. Ten patients were mechanically ventilated due to ALI (ALI group), which was defined according to the re cent definitions for ALP Meas ure ments in ALI patients were performed be tween lO and 36 h after admission to the ICU. Nine patients had a longstanding histmy of chronic obstructive lung disease (COPD

group). The diagnosis was based on clinical examination and previous pulmonary function test results from their medical records. Four of the COPD patients were mechanically ventilated postoperatively and studied within 4 h fater surgery. In five COPD patients, mechanical ventilation was commenced due to respiratory insufficiency caused by bronchopulmonary infection. These five patients were studied within 72 h after onset of mechanical ventilation. COPD patients were treated with bronchodilators such as 13 2-adrenergic agonists, theophylline, or a combination of both. The demographic data and entry criteria of all pati en ts are summarized in Table l. Prior to meas urements, the ventilator was adjusted (VT, respiratmy rate, Fio 2 , PEEP) according to clinical needs, then PEEP was decreased to 0 em H 2 0 , and occasionally Fio 2 was adapted to match th e previous arterial oxygen saturation. The EELV was changed by increasing PEEP from 0 to 5 to lO em H 2 0 and thereafte r declining to 0 em H 2 0 again. EELV was determined by N2 WO at the beginning of the experiment with PEEP of 0 em H 2 0 and thereafter at each PEEP level. After each N 2WO, a nitrogen wash-in procedure was performed by switching the Fio 2 back to baseline. One N2 WO followed by a nitrogen wash-in ( patient) to 15 min (COPD patient). Thereafter, required 6 ALI PEEP was changed and an additional 5 min were waited before a new N2 WO was started, resulting in a total study period of 45 to 80 min in the individual patient. RIP was performed continuously with a respiratory monitor (Respitrace Plus; NIMS Inc; Miami ) operating in the DC mode. The RIP signal was recorded with a personal computer using an analog-digital converter (DT 2801-A; Data Translation; Marlborough, Mass) . Within the first 5 min of operation, the Respitrace respiratory monitor performs an autocalibration as described in detail by Sackner and coworkers 7 to determine the volume motion coefficients of the rib cage and abdominal cage coils. During th e calibration and the following study pe riod, all patients remained unchanged in th e supine position. The rib cage and abdominal cage coils were aligned with the nipples and umbilicus, re spectively, and tightly fixated but not taped to the naked body as recom mended by the manufacturer. A quantitative calibration of the RIP signal was done off-line during the first minute of controlled mechanical ventilation. The RIP sum signal was compared with the simultaneously recorded volume signal obtained from the integrated gas flow signal that was measured by PT. Gas flow was measured with a heated pneumotachograph

Table ! - Demographic Characteristics and Entry Criteria of Patients* Group No. (F/1vl) Diagnosis

Age, yr Height, em Weight, kg Fio2 Pa0 2 (Fio 2 = l.O), mm Hg FRC, mL Pawmean, em H 2 0 Cst, mUcm H 2 0 Raw, em H 2 0 s/L

12 (7/5) Abdominal surgery Hip or knee replacement 49.6::'::17.1 169::'::11 67::'::15 0.34::'::0.1 470::':47 2,256::':758 7.6::':1.5 61.4::':11.8 12.3::'::3. 1

COPD

ALI

PO STOP (10) (2)

10 (3n) Lung contusion Sepsis Pneumonia 53.1::'::20.0 172::'::9 70::'::16 0.46::':8 232::':38 1,351::': 313 9.3::':1.1 43.1::':10.8 13.3::':2.7

(3) (4) (3)

9 (0/9) Pul monary infection Neurosurgery Osteosynthesis 68.7::'::9.6 177::'::4 81::'::9 0.41 ::':8 354::':96 3,198::': 1,299 8.8::': 1.4 60.1::':27.4 16.2::':6.6

(5) (3) (1)

*Pawmean=mean airway pressure; Pawmean>Cst, and Raw were measured at the beginning of the study with PEEP=O em H2 0. Numbers in parentheses behind diagnoses are the number of p ati ents. Data are given as mean ::':SD. Note that we did not measure intrinsic PEEP. Therefore th e Cst, especially in COPD patients, might be an underestimation of the true value. 444

Clinical Investigations in Critical Care

(Fleisch No. 2; Lausanne, Switzerland) inserted between the endotracheal tube and the Y-piece of the ventilatory circuit and a differential pressure transducer (H uba Control; Wi.irenlos, Switzerland ). The response of the PT was linear over the experimental flow range (0 to 1.5 U s). The flow-m easuring system w as calibrated with a gas mixture with known gas concentrations and viscosity using a precision calibration pump (Engstrom Megamed 05; Stockholm , Sweden) producing a sinusoidal flow patte rn. The instantaneous gas viscosity was dete rmined from the analyzed gas fractions to correct the measured flow signal. 8 Volume was then obtained from the corrected flow signal during off-line analysis. To minimize a drift of the volume signal by an off-set of the flow signal, the pressure transducer was adjusted m eti culously during zero-flow conditions before the measurement. Furthermore, the pressure transducer was automatically readjusted during the first second of every minute of the data acquisition period by a n automatic ze roing procedure that is built into our custom-designed m easuring system. Thus, the PT signal showed no appreciable shift which would have easily been r ecognized during zero-flow periods. Airway pressure was measured b etwee n the endotracheal tube and the PT with a differential pressure transducer (Huba Control) that was calibrated prior to each measurement with a precision calibration pump (Engstrom Megamed 05). Quasi static compliance (Cst) and airway r esistance (Raw) were calculated according to standard formulas: Cst=VT/ (Pplateau- Pend-exp; mto•y), Raw= (P pe ak - Ppiateau)/flow. Gas concentrations were measured with a mass spectrometer (Perkin Elmer 1100 A; Medical Gas Analyzer; Pomona, Calif). Signals of flow and gas concentration were synchroni zed during off-line analysis to compensate the gas sampling and data processing delay of the mass spectrometer. Furthermore, mathematical corrections of viscosity changes have been included due to the gas concentration variation during the washout maneuver. 8 EELV was determined b y N2 WO. The inspiratory oxygen concentration was changed to 100% and a mass balance of inspired and expired nitrogen was calculated using integrated nitrogen concentration and flow signals. The washout maneuver was performed until the end-expiratory nitrogen concentration was < 1.5%. Nitrogen w ashed out from the body tissues was taken into account by formulas from the literature 9 •10 adapted to the patient's body surface area. In preliminary studies in a lung model, the accuracy of this method was ±5% for lung model volumes between 500 and 7,000 mL. During duplicate measurements in patients, the mean difference was 1.15±3.1 %. The baseline drift of the RIP signal was determined three times for each patient by comparing the EELV over 5-min periods of stable volume-controlled v entilation between minute 1to 10, 20 to 30, and 40 to 50. For comparison of VT between RIP and PT, a quantitative calibration of the RIP signal was performed only once at minute l. Then the data were analyzed during consecutive minutes asthe mean difference per minute between RIP and PT volume measurements (difference [%]=[RIP- PT]/PT] X 100). To estimate the effect of the baseline drift on LlEELV determination with RIP, three different evaluation procedures were performed in all patients. Calculation 1

LlEELV was determined for a 1-min period by comparing the end-expiratory RIP signal immediately before and approximately 50s after the change of PEEP (Fig 1). This short time period was chosen, as Katz e t alLI could demonstrate that in patients with acute pulmonary failure, PEEP-induced LlEELV is 90% com-

Baseline Drift and Delta-EELV Detennination of RIP PEEP increase of 5 cl'l1lli0

~

Tinle FIGURE l. Baseline drift of the RIP sum signal was determined during stable volume-controlled v entilation. LlEELV induced by the application of PEEP erroneously includes the baseline drift that occurs during this time period of measuring.

plete already after 3 to 5 breaths. Thus, we hypothesized that this short time period might minimize the influence of the baseline drift. Calculation 2

LlEELV was determined as described above with res ults being corrected for the baseline drift that had occurred within the last minute before the PEEP changed. Calculation 3

LlEELV was measured over a 5-min period in order to detect the entire PEEP effect on EELV at the moment when the N2 WO started. This LlEELV was compensated for the baseline drift that had occurred within the last 5 min before the PEEP had changed. The LlEELV determined by RIP with the three different calculation procedures was compared to the corresponding LlEELV determined by N 2 WO. All data are g vi en as means±SD. For comparison of subgroups, the two-tailed Student's t test was applied, with a l evel of significance at p < 0.05.

RESULTS

Analysis of the baseline drift of the RIP signal (Fig 2) revealed an average increase of +25.4:±:29.1 mUmin with 28.2:±:48.9, 24.5:±:16.6, and 23.5:±:20.7 mUmin for the first , second, and third 5-min period, respectively. In 51% of all minutes analyzed, drift was in the range of 0 to +20 mUmin and in 98%, it was in the range of -20 to +80 mUmin. However, in individual patients, the drift was neither stable nor steadily increasing or decreasing v.rith time, but changed over a wide range as indicated in Figure 3. The comparison of VT for each minute between RIP and PT (Fig 4) revealed that 66.5% of all values were :±: 10% accurate and 90.0% of all values w ere :±:20% accurate . The deviation between RIP and PT CHEST I 113 I 2 I FEBRUARY, 1998

445

Baseline Drift of the RIP-Signal in aU Patients

0--20 - 10

0

10

20

30 40 50 [ml / min]

60

70

80

272

F IGURE 2. The abscissa shows th e baseline drift (mUmin ); the ordinate gives th e percentage of data within the indicated margins (min < X::S max).

VT measurements during the study period did not increase with the tim e period elapsed after calibration. However, there were significant differences among the three groups of patients (Table 2). Measurements in COPD patients differed more between both methods ( - 9.3 ±13.9%, p < 0.01 ) than in patients with ALI ( - 2.9 ±9.9%) or normal pulmonary function (-2. 1±10.6%). The difference between ALI and POSTOP patients was also significant (p<0.05 ). In the ALI group, less values were within the range of ± 10% but more values were within ± 20% (Fig 5). Mean EELV of all patients determined b y N2 WO was 2,189± 1,084 mL at PE EP of 0 em H 2 0 , in-

creased to 2,441 ± 1,177 mL and 2,881 ± 1,212 mL at PEEP levels of 5 em H 2 0 and 10 em H 2 0 , respectively, and decreased to 2,232±1,121 mL after changing back to PEEP of 0 em H 2 0 again. EELV was lowest in ALI patients (p < 0.01 ) and highest in COPD patients. The differences between POSTOP and COPD patients was not signiflcant (p<0.2) (Fig 6). Linear regression of AEELV between RIP and N2 WO measurements was best for the group of ALI patients with correlation coefficients of 0.86, 0.87, and 0.82 for the calculation procedures 1, 2,and 3, respectively. For POSTOP patients, these correlation coefficients are 0.77, 0.77, and 0.71 and for the COPD patients, they are 0.77, 0.72, and 0.59. Thus, the results with the calculation procedures 1 and 2 (evaluation of a 1-min pe1iod) were just slightly different and better than tl1e results with procedure 3 (measuring a 5 -min period). The equations of the regression lines for calculation procedure 1 are AEELVRIP=0.95X AEELVN 2 w o+O for ALI patients, AEELVRIP= 0.95XAEELVN2 w o -8 for POSTOP patients, and AEELVRIP=0.59 XAEELVN2 wo + 193 for COPD patients. For all data, the difference of AEELV between botl1 methods was 11.7±174.1 mL using procedure 1 (Fig 7). However, compared to N2WO, only 25.8%, 46.2%, 100%, with eight of the nine measurements being done in COPD patients.

Baseline Drift of the RIP-Signal in Individual Measurements

[ml/min]

100

272

80 N = 12

60

N= 10

N=9

40 20 0 -20

f-

- - r-

r oI

nJ~

POSTOP

~ m,J H ~ ~~

::

I

I

I

i-

J1 I

ALI

<

: :·

~:

n ~: .

I

r-

r 1--

~w I

I

-

J

..:· : :

:

:

COPD

FIGURE 3. Each section of th e abscissa shows three determinations of the baseline drift (mUmin) averaged over 5-min periods dming stable volume-controlled ventilation in indivi dual patients. The first bar of each section i s the baseline drift determined within the flrst 10 min, the second bar is between minute 21 and 30, and the third bar is between minute 41 and 50. Section 2 includes only two bars because baseline dtift was 0 within the third 5-min period.

446

Clinical Investigations in C ritical Care

Difference ofVT during consecutive minutes 50,-~--~------~--~~--~--~~

30

:C Min-Max

-50

. . 25%-75% 5

10

15

20

25

30

35

40

45

D

50

Median value

Minutes FIGURE 4.

minute l.

Each minute includes data from all recorded measurements. Calibration was done in

DISCUSSION

Our data show that RIP is not consistently precise enough (deviation <10%) for quantitative measurements ofVT in mechanically ventilated patients. The difference between N 2 WO and RIP for LlEELV was even greater than between PT and RIP for VT measurements. However, this might also partially reflect problems of N 2 WO to measure the complete intrathoracic gas volume. Using the Respitrace Plus monitor, we found a high and unstable baseline drift of the RIP signal that certainly hindered a higher accuracy of this method, especially for long-term measurements. RIP detects thoracic gas volume changes in COPD

patients less reliably than in ALI or patients with normal pulmonary function. The average baseline drift of the RIP signal in our study is 25.4:±:29.1 mUmin, which is considerably higher than data reported in the literature: 1 mU min, 12 <8.5 mUmin,U <10 mUmin,l4 and 187:±:69 mU120 min. 15 This drift cannot be attributed to a gradual displacement of the rib cage and abdominal cage coil, since this would result most likely in less stretching of the coils and a "negative drift." Even more important, in the present study, the drift changed rapidly and over a wide range in individual patients with a maximum of 1,360 mL for a 5-min period (ie, 272 mUmin), and it was therefore unpre-

Table 2-Accuracy ofVr Determination*

Mean deviation, % Accuracy :':: 10%, % Accuracy :'::20%, % Range of deviation , %

All Patients

PO STOP

ALI

COPD

-4.5:'::11.9 66.5 90.0 96.9

-2.1:'::10.6 72.1 91.1 74.5

-2.9:'::9.9 67.1 95.3 75.2

-9.3:':: 13.9 60.3 83.7 78.4

*For abbreviations see legend of Table 1 and text. VTs are compared for consecutive minutes. Difference (%=[RIP - PT]/PTX100). CHEST /113/2/ FEBRUARY, 1998

447

Difference of Minute Ventilation between RIP and PT in the Patient Groups 50 ,........,

~ ......... r:n

:::::: 0

·~

r:

vr:n

..0

0

45 40 35 30 25

ii POSTOP

20 v 15 """ 10 5

c.., 0

-a

z

•ALI

0 -70 -60 -50 -40 -30 -20 -10

0

10 20 30 Difference of Minute Ventilation [ % ]

40

DCOPD

FIGURE 5. VTS were calcu lated for consecutive minutes in each patient. The abscissa shows the difference of minute ventilation ( min < X~ max ) between RIP and PT. The ordinate gives the percentage of obse1vations in each subgroup.

dictable for a certain time pe1iod. Oscillator function is sensitive to temperature changes and thermal instability of the RIP device (ie, Respitrace Plus) has been demonstrated. 15 The high d1ift could therefore be du e to an insufficient acclimation period prior to th e beginning of the individual measurements . However, since we did not find a significant decrease of

the baseline drift with increasing time dming measurements, this explanation seems very unlikely. Ambient temperature changes in th e ICU can practically be neglected, leaving changes of the body temperature of patients as one possible reason. After major surge1y, many patients are admitted to the ICU with moderate hypothermia (34 to 35°C). Oc-

EELVin Patient Groups

s

..........

.........,

>

ffi

5000 4000 3000 2000

-+------==-

1000 0

PEEP

0 5 10 0

ALI

0 5 10 0

PO STOP

0 5 10 0

COPD

F I GURE 6. The EELV determined b y N2WO is shown as mean:tSD for each PEEP level in the three different patient groups.

448

Clinical Investigations in Critical Care

Comparison of Delta-EELV between N2WO and RIP

•

600 0

~

400

(+o<

0


t::


1-o

~

~

>

ffi

-

,........

200

8 ..........

~

'"0

~

-200

~ ~

- - - - - - - ~- - - - - - - - .. - - - - - - - - - - - -- 2 SD

-400

~·

I

cd

~


Q

-600

•

• ALI ~ COPD • POSTOP

-800+-----~----~----+-----+-----+-----+---~

0

200 400 600 800 1000 1200 Mean ofDelta-EELV between N 2WO and RIP [ml]

1400

FIGURE 7. Data of calculation procedure 1are shown according to Bland and Altman 18 for comparison of two different methods.

casionally, they increase with body temperature beyond 39°C before stabilizing at 37°C again. In patients with ALI due to an underlying inflammatory process, instability of the body temperature is also common. Thus, changes of the body temperature that are transmitted to the rib cage and abdominal cage coil might have contributed to the high and unpredictable baseline drift. Unfortunately, we did not measure skin temperature changes in our patients. This hypothetical effect should be investigated in further studies. The drift might also partially reflect changes in intrathoracic blood volume, since volume replacement was done according to clinical needs unaffected by the study. We found a good agreement between PT and RIP in the measurement of VT with 66.5% and 90% of all recorded values being ±10% and ±20%, respectively, which is almost identical to previously published results. 6•7 Baseline drift is obviously no problem for the determination ofVT due to the short duration (2 to 4 s) of the inspiratory and expiratory cycle of a single breath. VT determination by RIP was significantly less accurate (p<0.01) in COPD patients compared to the other patients. In COPD patients, the mean difference of VT between PT and RIP was 9.3± 13.9% and only 83% of all data were accurate within ±20%. Similar results in COPD patients were previously reported by Hudgel and coworkers 13 who demonstrated a mean error of 7.6%. Sixty-seven

percent of their data were precise within the range of ±20% in this group of patients. A slightly higher deviation of VT determined by RIP in COPD patients compared to normal patients is also reported by Gonzalez and coworkers.l 6 In contrast, Tobin and coworkers,l7 using the Respitrace respiratory monitor, found no relationship between RIP deviation and severity of airway obstruction, but they used only a short validation period of "at least 20 seconds" and the rib cage and abdominal cage coil were adjusted with a new calibration performed if a deviation of > 10% was detected in their study. Difficulties of RIP to detect thoracic volume changes in COPD patients have been attributed to additional degrees of freedom of the chest wall, 16· 17 which might not be detected by the rib cage and abdominal cage coil. For determination of 11EELV, the overall accuracy of RIP in regard to N2 WO seems acceptable considering correlation coefficients between 0.77 and 0.86 for calculation procedure 1 and an average deviation of 11.7±174.1 mL. However, when data are plotted as suggested by Bland and Altman 18 for comparison of the same parameter by two different methods (Fig 7), it is obvious that the difference of 11EELV between N 2 WO and RIP is unacceptably high (> 10%) in many cases. Just 25.8% of all values were accurate ± 10%. This is similar to the data presented by Werchowski and coworkers 6 (30%±10% accurate) for spontaneously breathing patients and EELV changes induced by continuous positive airway presCHEST I 113 I 2 I FEBRUARY, 1998

449

sure (CPAP). In contrast, Sackner and coworkers 7 demonstrated that 80% of all EELV changes were detected by RIP with an accuracy of ± 10%. This difference might be related to the method by which EELV was raised. We as well as Werchowski and coworkers6 used CPAP to increase EELV, while in the study of Sackner and coworkers, 7 EELV was voluntarily increased during spontaneous breathing. A higher intrathoracic pressure as induced by CPAP, however, results in an increased functional residual capacity (FRC) but decreases simultaneously the intrathoracic blood volume. 19 This effect is most pronounced in hypovolemic subjects 20 and might lead to an underestimation of the FRC as determined by RIP. Vice versa, saline solution infusion of 30 mUkg has been shown to decrease the FRC by 10%,2 1 which would not be detected by RIP. One study compared the thoracic gas volume in awake and anesthetized subjects measured by body plethysmography with the circumference and the anterior-posterior diameter of the rib cage and abdomen. Only minor and nonsignificant changes of the thoracic and abdominal dimensions were detected after induction of anesthesia, which could not explain the decrease of thoracic gas volume observed. 22 The dif~ ferences between the FRC measured by RIP and N2 WO might therefore be to some extent caused by the counterbalance of the intrathoracic gas and blood volume. However, this counterbalance should lead to a constant underestimation of the ~EELVRIP, which is not the case in the present study (Fig 7). Since our intention was to investigate the ability and accuracy of RIP for the determination of FRC in a clinical setting, we did not control the administration of fluids during the course of the study. This was done merely by clinical needs and might have contiibuted to the differing results of both methods. However, all patients were in hemodynamically stable condition and PEEP increases induced only minor declines of the BP, thus, marked hypovolemia was not present in our patients. The unexpectedly high baseline drift of the RIP signal has also obviously hindered a more precise ~EELV determination. The influence of the baseline drift increases with the duration of ~EELV measurement. Thus, calculation procedure 3 with a 5-min evaluation of PEEP-induced ~EELV was less accurate than calculation 1 and 2 with a 1-min evaluation each. Therefore, we believe that measuring ~EELV by RIP could be improved if a better compensation of the baseline drift is achieved. ~EELV determination by RIP and N 2WO differed most in COPD patients. Linear regression analysis between RIP and N 2 WO ~EELV revealed a slope close to 1 and an ordinate intercept close to 0 mL for ALI and POSTOP patients (procedure 1), while in the COPD group, the slope was 0.59 with an 450

intercept of 193 mL. This indicates a systematic difference between the two methods in these patients. Furthermore, all ~EELV determinations with a difference > 100% between both methods were found in the COPD group, except for one single value of a POSTOP patient. N 2 WO measures only the gas volume, which is ventilated during the washout procedure. This volume has therefore been called "accessible pulmonary volume" 23 and is not necessarily equal to the FRC. Trapped air caused by mucous plugs or airway closure is not detected by N 2WO and the degree of ainvay closure might decrease as a result of PEEP. This would lead to an overestimation of ~EELVN 2 wo with increasing PEEP levels and vice versa. Incomplete recove1y of nitrogen due to inhomogeneous ventilation24 and gas mixing inefficiency2.s·26 is another potential source of error in COPD patients which might cause an underestimation of the FRC determined by N2 WO. In spontaneously breathing COPD patients vvith respiratmy failure and intrinsic PEEP, extrinsic PEEP has been reported to improve gas exchange without altering EELV; 27 thus, an influence of PEEP on gas mixing and ventilation inhomogeneity must be considered that is not accompanied by a change of the FRC, but has the potential to influence the results of N2WO. We did not measure intrinsic PEEP but seven of nine COPD patients had some end-expiratory residual flow indicating intrinsic PEEP during ventilation with PEEP 0. Since the application of PEEP 5 caused an FRC increase in eight of nine COPD patients that averaged 204 mL (201 mL ALI, 297 mL POSTOP), we believe that intrinsic PEEP alone does not explain our differing results. In addition, the study of Goldberg and coworkers 27 used RIP for ~EELV determination which might have failed to detect subtle changes of ~EELV. Actually the authors stated that the measured increase of the esophageal pressure should have amounted to an increase of approximately 200 mL in lung volume that was obviously not detected by RIP. In controlled mechanically ventilated patients with severe COPD, PEEP improved gas exchange only parallel to an increase of EELV.28 Nitrogen washed out of the blood and body tissue influences the results of N2WO and might have a greater impact on the results in COPD patients because of the longer time periods required to reach an end-tidal N2 concentration of 1.5%. In our FRC calculation procedure, this effect was minimized by a mathematical correction taking the time of the N 2WO maneuver and the patient's body weight and height into account. 9 ·l0 The above-mentioned problems ofN 2WO to measure precisely changes ofEELV, especially in COPD patients, must be kept in mind when interpreting the results. Clinical Investigations in Critical Care

However, as pointed out above also, RIP has considerable problems regarding volume measurements in this group of patients. The assumption that the chest wall has two degrees of freedom during open breathing, 29 which is essential for volume measurements with RIP, seems to be questionable for COPD patients. 16 ,17 In conclusion, we did not find RIP to be consistently precise enough for quantitative measurements of VT and our results indicate that this is most likely true for the determination of ~EELV as well. With the Respitrace Plus respiratory monitor, the high and unpredictable baseline drift in the DC mode jeopardizes a more accurate ~EELV determination. However, RIP can surely be used when only semiquantitative data are required as for optimizing ventilator settings in the ICU. The difference of RIP and N 2 WO to detect thoracic gas volume changes was greater in patients with COPD compared to patients with ALI or normal pulmonary function .

12

13 14 15

16

17 18

REFERENCES 1 Dall'Ava Santucci J, Armaganidis A. Respiratmy inductive plethysmography. In: Benito S, Net A, eds. Pulmonary function in mechanically ventilated patients: update in intensive care and emergency medicine. New York: Springer, 1991; 121-42 2 Langenstein H, Brunner J, Wolff G. Zum verhalten der funktion ellen residualkapazitat bei akuter respiratorischer insuffizienz. Schweiz Med Wochenschr 1983; 113:1141-44 3 Bernard G, Artigas A, Brigham K, e t a!. Report of the American-European consensus conference on ARDS: definitions, mechanisms, relevant outcomes and clinical trial coordination. Intensive Care Med 1994; 20:225-32 4 Cole A, Weiler S, Sykes M. Inverse ratio ventilation compared with PEEP in adult respirato1y failure. Intensive Care Med 1984; 10:227-32 5 Spence D, Kelly Y, Ahmed J, et a!. Critical evaluation of computerized x ray planimetry for the measurement of lung volumes. Thorax 1995; 50:383-86 6 Werchowski JL, Sanders MH , Costantino JP, et al. Inductance plethysmograph measurement of CPAP-induced changes in end-expiratory lung volume. J Appl Physiol 1990; 68:1732-38 7 Sackner MA, Watson H , Belsito AS, et a!. Calibration of respiratory inductive plethysmograph during natural breathing. J Appl Physiol 1989; 66:410-20 8 Brunner J, Langenstein H, Wolff G. Direct accurate gas flow measurement in the patient: compensation for unavoidable error. Med Prog Techno! 1983; 9:233-38 9 Cumming G. Gas mixing efficiency in the human lung. Respir Physiol 1967; 2:213-24 10 Cournand A, Yarmouth IG, Riley RL. Influence of body size on gaseous nitrogen elimination during high oxygen breathing. Proc Soc Exp Bioi 1941; 48:280-84 11 Katz J,Ozanne G, Zinn S, eta!. Time course and mechanisms

19

20 21 22 23 24 25 26

27 28

29

of lung-volume increase with PEEP in acute pulmonary failure. Anesthesiology 1981; 54:9-16 Yalta P, Takala J, Foster R, eta!. Evaluation of respiratory inductive plethysmography in the measurement of breathing pattern and PEEP-induced changes in lung volume. Chest 1992; 102:234-38 Hudgel DW, Capehart M, Johnson B, eta!. Accuracy of tidal volume, lung volume and flow measurement by inductance vest in COPD patients. J Appl Physiol 1984; 56:1659-65 Dall'Ava-Santucci J, Armaganidis A, Brunet F ,eta!. Causes of error of respiratory pressure-volume curves in paralyzed subjects. J Appl Physiol 1988; 64:42-49 Nunes J, Leino K, Yalta P, et a!. Evaluation of a new respiratory inductive plethysmograph in the measurement of PEEP induced changes in lung volume [abstract]. Intensive Care Med 1995; 21(suppl 1):S51 Gonzalez H, Haller B, Watson H , et a!. Accuracy of respiratory inductive plethysmograph over wide range of rib cage and abdominal compartmental contributions to tidal volume in normal subjects and in patients with chronic obstructive pulmonary disease. Am Rev Respir Dis 1984; 130:171-74 Tobin MJ, Jenouri G, Lind B, et al. Validation of respiratory inductive plethysmography in patients with pulmonary disease. Chest 1983; 83:615-20 Bland JM , Altman DC. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-10 Nielsen J, Sjostrand U, Edgren E , et a!. An experimental study of different ventilatory modes in piglets in severe respiratory distress induced by surfactant depletion. Intensive Care Med 1991; 17:225-33 Lichtwarck-Aschoff M. Intrathoracic fluid and gas compartments with different modes of ventilation [thesis]. Acta Univ Upsaliensis 1994; 447:34-56 Goeree G, Coates G, Powles A, etal. Mechanisms of changes in nitrogen washout and lung volumes after saline infusion in humans. Am Rev Respir Dis 1987; 136:824-28 Hedenstierna G, LOfstrom B, Lundh R. Thoracic gas volume and chest-abdomen dimensions during anesthesia and muscle paralysis. Anesthesiology 1981; 55:499-506 Brunner J, Wolff G. Pulmonary function indices in critical care patients. New York: Springer-Verlag, 1988 Guisan M, Tisi GM, Ashburn WL, et a!. Washout of 133xenon gas from the lungs: comparison with nitrogen washout. Chest 1972; 62:146-51 Cumming G, Guyatt R. Alveolar gas mixing efficiency in the human lung. Clin Sci 1982; 62:541-47 Fleming G, Chester E, Saniie J,et a!. Ventilation inhomogeneity using multibreath nitrogen washout: comparison of moment ratios and other indexes. Am Rev Respir Dis 1980; 121:789-94 Goldberg P, Reissmann H, Maltais F, et a!. Efficacy of noninvasive CPAP in COPD with acute respiratory failure. Eur Respir J 1995; 8:1894-1900 Ranieri M, Giuliani R, Cinnella G, eta!. Physiologic effects of positive-expiratory pressure in patients with chronic obstructive pulmona1y disease during acute ventilatory failure and controlled mechanical ventilation. Am Rev Respir Dis 1993; 147:5-13 Konno K, Mead J. Measurement of the separate volume changes of rib cage and abdomen during breathing. J Appl Physiol1967; 22:407-22

CHEST I 113 I 2 I FEBRUARY, 1998

451