Accurate measurement of pulmonary ventilation. Tidal Volume and Respiratory Timing Derived From a Portable Ventilation Monitor*

Transcription

1 Tidal Volume and Respiratory Timing Derived From a Portable Ventilation Monitor* F. Dennis McCool, MD, FCCP; John Wang, MD; and Kristi L. Ebi, PhD Study objectives: To determine the accuracy of a portable magnetometer designed to measure tidal volume (VT), inspiratory time (TI), and expiratory time (TE). Participants: Fourteen healthy subjects. Design: Subjects breathed over a sixfold range of VTs while at rest (sitting and standing) and during treadmill exercise. We then compared VT,TI, and TE measured by magnetometry (VTmag, TImag, and TEmag) with VT, TI, and TE measured by spirometry (VTspiro, TIspiro, and TEspiro, respectively). Setting: Pulmonary function and exercise physiology laboratories. Measurements and results: The sternal-umbilical distance and the anteroposterior displacements of the rib cage and abdomen were measured with two pairs of magnetometer coils. VT was calculated from the sum of these three signals, and was simultaneously measured using a spirometer or flow meter. A total of 1,111 breaths were analyzed for the resting condition, and 1,163 breaths were analyzed for the exercise condition. We found that VTmag was highly correlated with VTspiro at rest (R ) and during exercise (R ) for pooled data. The slope of this relationship approached the line of identity. The mean percentage differences between VTmag and VTspiro were % at rest and % with exercise. By Bland-Altman analysis, the mean differences between VTmag and VTspiro were 38 ml at rest with changes in posture, and 182 ml during exercise. TImag and TIspiro values and TEmag and TEspiro values also were highly correlated (R and R , respectively) for pooled data. Conclusion: A portable magnetometer system can give useful measures of VT, TI, and TE over a wide range of VTs in sitting, standing, and exercising subjects. (CHEST 2002; 122: ) Key words: pulmonary ventilation; thorax; tidal volume Abbreviations: RIP respiratory inductive plethysmography; Te expiratory time; Temag expiratory time measured by magnetometry; Tespiro expiratory time measured by spirometry; Ti inspiratory time; Timag inspiratory time measured by magnetometry; Tispiro inspiratory time measured by spirometry; V e minute ventilation; Vt tidal volume; Vtmag tidal volume measured by magnetometry; Vtsit tidal volume measured while sitting; Vtspiro tidal volume measured by spirometry; Vtstd tidal volume measured while standing; Xi sternal-umbilical distance Accurate measurement of pulmonary ventilation requires the use of devices such as masks or mouthpieces coupled to the airway opening. Unfortunately, these devices are both encumbering and invasive, and thus ill suited for measurements in the ambulatory setting. Alternatively, devices that sense *From the Departments of Medicine, Memorial Hospital of Rhode Island, and Brown University Medical School, Pawtucket, RI. Supported by a contract with the Electric Power Research Institute. Manuscript received July 31, 2001; revision accepted February 6, Correspondence to: F. Dennis McCool, MD, FCCP, Memorial Hospital of Rhode Island, 111 Brewster St, Pawtucket, RI 02809; F_McCool@Brown.edu respiratory excursions at the body surface can be used to measure pulmonary ventilation. Konno and Mead 1 extensively evaluated a two-degrees-of-freedom model of chest wall motion, whereby ventilation could be derived from measurements of rib cage and abdomen displacements. With this model, tidal volume (Vt) was calculated as the sum of the anteroposterior dimensions of the rib cage and abdomen, and could be measured to within 10% of actual Vt as long as a given posture was maintained. Smith and Mead 2 subsequently found that changes in posture associated with movements of the spine and pelvis caused large displacements of the chest wall. This axial displacement of the chest wall 684 Clinical Investigations in Critical Care

2 comprised a third degree of freedom of chest wall motion that could be assessed by measuring the distance between the xiphi-sternal junction and pubic symphysis. McCool et al 3 then used a threedegrees-of-freedom model to calculate Vt as the sum of this axial measure of chest wall motion and the motion of the rib cage and abdomen. Using this approach, Vt was accurately assessed in subjects performing varied postural activities. 3,4 McCool and Paek 5 then used a three-degrees-offreedom approach to measure ventilation in a work environment outside the laboratory. This was accomplished by using numerous pieces of bulky equipment powered from wall outlets. These methods were encumbering for both subjects and investigators. To avert these shortcomings and design a device more suitable for field studies, we developed a magnetometer appliance (Enertech Consultants; Campbell, CA) that is small, portable, lightweight, and powered by batteries. The purpose of the present study was to determine the accuracy of this device in measuring Vt, inspiratory time (Ti), and expiratory time (Te), while subjects were seated, standing, or performing exercise on a treadmill. Materials and Methods measured using two pairs of one-half inch in diameter electromagnetic coils secured to the rib cage and abdomen. Cables connected the coils to the magnetometer device, which measures cm, weighs 430 g, and is powered by a battery pack. The device consists of a flowmeter, transmitter, and receiver circuitry. The two transmitter coils are connected to oscillators set at 8.97 KHz and 7.0 KHz, respectively (Fig 1). The 8.97-KHz transmitter coil is placed in a posterior position over the spine at the midsternal level. The 7.0-KHz transmitter coil is placed anteriorly in the midline of the abdominal wall just above the umbilicus. One receiver coil is tuned to 7.0 KHz and placed in a posterior position over the spine at the level of the umbilicus. This coil acts as a receiver for the abdomen anteroposterior dimension. The second receiver coil is tuned to both the 7.0-KHz and 8.97-KHz frequencies. The second coil acts as a receiver for both the sternal-umbilical signal and the rib cage anteroposterior signal. The received signals are processed as three channels by three detection circuits. Dual functionality of the second receiver coil eliminates the need for a third pair of coils, thereby simplifying design and decreasing power needs. Protocol Each subject performed calibration maneuvers as previously described. 3,4 Briefly, this maneuver consisted of taking 15 to 20 breaths of varied depth through a flowmeter incorporated into the magnetometer device. The subjects were instructed to take breaths while primarily using the rib cage, then while primarily using the abdomen, and finally while flexing and extending the spine. Calibration maneuvers were performed while the subject was sitting and standing. After calibration, nine subjects were studied at rest while Subjects Fourteen healthy male and female subjects were recruited. A total of nine subjects participated in the studies performed at rest, and nine subjects participated in the exercise studies (Table 1). Institutional review board approval and informed consent were obtained. Measurements The anteroposterior displacements of the rib cage and abdomen, as well as the axial displacements of the chest wall, were Table 1 Subject Characteristics Subject No. Age, yr Height, cm Weight, kg Figure 1. Schematic of the magnetometer device depicting coil placement. The transmitter coils (yellow and black) oscillate at 8.97 KHz and 7.0 KHz, respectively. The red receiver coil is tuned to both frequencies, and the green receiver coil is tuned only to 7.0 KHz. CHEST / 122 / 2/ AUGUST,

3 seated and standing. They were instructed to take breaths ranging from 0.5 to 3.5 L while breathing into a spirometer (Transfer Test Model C; PK Morgan; Chatham, Kent, England) for a total of 6 min. For the exercise condition, nine subjects exercised on a treadmill for 6 min while breathing through a flowmeter in-line with a metabolic cart (Sensormedics, Yorba Linda, CA). The duration of exercise was 6 min. The treadmill was set at a maximal incline of 3 and a maximal speed of 3.5 miles per hour. At this speed, subjects were walking at a brisk pace and sweating. Calculations The volume of air inhaled and exhaled was calculated as the sum of the changes in the anteroposterior distance of the rib cage, the anteroposterior distance of the abdomen, and the axial displacement of the chest wall (the sternal-umbilical distance [Xi]): Vt RC Ab Xi, where,, and are coefficients for the rib cage (RC), the abdomen (Ab), and the Xi. The values of the coefficients were determined for both the sitting and standing positions by multiple linear regression of the flowmeter-derived volume with the anteroposterior displacements of the rib cage and abdomen and the axial displacement of the chest wall. The coefficient of determination of the actual vs magnetometer-derived volume was calculated. If the coefficient of determination was 0.85, the calibration maneuver was reanalyzed. The calibration process needed to be repeated 3% of the time. The coefficients were then applied to magnetometer signals for breaths taken in the corresponding sitting or standing positions. In addition, coefficients for the sitting position were applied to magnetometer signals obtained while the subject was standing. Analysis Vt, Ti, and Te measured by magnetometry (Vtmag, Timag, and Temag, respectively) were regressed against Vt, Ti, and Te measured by spirometry (Vtspiro, Tispiro, and Tespiro, respectively). Individual and pooled data were analyzed. The coefficient of determination (R 2 ), slope, and intercept were calculated for each parameter using a linear model. The mean percentage difference between the actual and magnetometer-derived breathing pattern parameters was calculated as the absolute value of (1 [magnetometry/spirometry]) 100. The bias (mean difference), estimated limits of agreement, and precision for Vt, Ti, and Te were calculated using the methods of Bland and Altman. 6 Results The pooled data for Vt, Ti, and Te during rest and exercise are shown in Figures 2, 3. A total of 1,111 breaths were analyzed over a sixfold range of Vts for the resting condition, and 1,163 breaths Figure 2. Pooled data for all subjects during resting breathing in the seated and standing positions. There were significant correlations between Vtmag and Vtspiro (R ; top, A); Timag and Tispiro (R ; middle, B); and Temag and Tespiro (R ; bottom, C). The slopes of each relationship approached the line of identity. sec seconds. 686 Clinical Investigations in Critical Care

4 Figure 3. Pooled data for all subjects during exercise. There was a significant correlation between Vtmag and Vtspiro (R ). The slope of this relationship approached the line of identity. were analyzed, also over a sixfold range of Vts, for the exercise condition. For pooled data obtained at rest with the subjects sitting and standing (Fig 2, top, A), Vtmag was highly correlated with Vtspiro (R ); the slope of this relationship approached the line of identity (Vtspiro 0.83 Vtmag 270). The results for Ti and Te were similar (R and R , respectively), with slopes also approaching the line of identity (Tispiro 0.99 Timag 0.01 and Tespiro 0.96 Temag 0.02; Fig 2, middle, B, and bottom, C). For pooled data during exercise, Vtmag was again highly correlated with Vtspiro (R ; Fig 3). The slope of this relationship also approached the line of identity (Vtspiro 0.89 Vtmag 370). For both the rest and exercise conditions, the magnetometer was less accurate for Vts 2,500 ml; Vtmag generally overestimated Vtspiro (Vtspiro 0.37 Vtmag). Coefficients of determination for Vt, Ti, and Te for breaths measured during rest and exercise were calculated for each individual. The mean R 2 value among subjects for Vt measured at rest while sitting (Vtsit) was 0.92, for Vt measured at rest while standing (Vtstd) was 0.94, for Vt measured at rest when applying the seated coefficients to the standing data (Vtsit Vtstd) was 0.91, and Vt while exercising was The values of R 2 for Vt ranged from 0.80 to The values of R 2 for Vt were generally lower when applying the sitting volume-motion coefficients to data obtained in the standing position (comparing Vtstd with Vtsit Vtstd). The mean R 2 among subjects for Ti measured at rest while sitting was 0.89, and for Ti measured at rest while standing was The mean R 2 among subjects for Te measured at rest while sitting was 0.90 at rest, and for Te measured at rest while standing was The mean percentage difference between the actual and magnetometer-derived values for Vt for each subject are shown in Table 2. The mean percentage differences for the group were % for the resting conditions (combined sitting and standing) and % for exercise. The two subjects with the greatest percentage error for Vt at rest (subjects 6 and 9) performed calibration maneuvers that were not as accurate in predicting actual Vt (lower R 2 ) as the other subjects. Similarly, the three subjects with the greatest percentage error for Vt during exercise (subjects 4, 11, and 13) performed less accurate calibration maneu- CHEST / 122 / 2/ AUGUST,

5 Table 2 Percentage of VTspiro vs VTmag Subject No. Mean Difference, % SD Exercise Mean Resting Mean vers. The mean percentage differences between the actual and magnetometer-derived values for Ti and Te for each subject are shown in Table 3. The mean percentage differences for the group were % for Ti and % for Te. The mean difference between Vtspiro and Vtmag (bias) and the 95% confidence intervals are plotted for Vt during changes in posture and Vt during exercise according to the methods of Bland and Altman (Fig 4). 6 The bias Vt at rest while sitting and standing was 38.2 ml, with a precision (SE) of 7.2 ml. The mean Vt for this data set was 1,350 ml. The bias Vt with exercise was ml, with a precision (SE) of 10.0 ml. The mean Vt for the exercise data set was 1,738 ml. The limits of agreement (95% confidence intervals) for Vt were narrower with the resting conditions ( 445 to 521 ml) Table 3 Percentage Differences of TIspiro and TImag Values and TEspiro and TEmag Values at Rest Subject No. Ti Difference, % SD Te Difference, % SD Mean than with exercise ( 505 to 869 ml; Table 4). The precision for the 95% confidence intervals at rest was 12.5 ml and during exercise was 17.4 ml. The best agreement between Vtspiro and Vtmag during exercise and at rest was over the lower range of Vts ( 2.5 L). The bias and limits of agreement for Vt with the resting conditions, Vt during exercise, Ti, and Te are summarized in Table 4. Discussion Changes in volume of the thoracic cavity can be inferred from displacements of the rib cage and diaphragm. Motion of the rib cage can be directly assessed, whereas the motion of the diaphragm is indirectly assessed as the outward movement of the anterolateral abdominal wall. Konno and Mead 1 used magnetometers to measure the displacements of the rib cage and abdomen and found that Vt can be calculated to within 10% of the Vt measured at the mouth when using the following two-degrees-offreedom model: Vt RC Ab, where RC and Ab represent displacements of the rib cage and abdomen, respectively, and and are coefficients. However, the two-degrees-of-freedom model is limited in accuracy when posture changes. First, the rib cage expands with isovolume spinal flexion or pelvic rotation. 2 As one bends forward, the abdominal contents displace the diaphragm cephalad, and this in turn expands the rib cage. 7 The overestimates in volume caused by this displacement error may be as great as 40 to 50% of the vital capacity. A second source of inaccuracy is caused by postural changes in volume-motion coefficients. 7,8 This error is of lesser magnitude than that related to rib cage displacement. Despite the above-mentioned limitations, the twodegrees-of-freedom model (equation 2) has been used extensively to monitor breathing in adults and children. Respiratory inductive plethysmography (RIP) is the most commonly used method using this model. As with magnetometry, RIP accurately measures Vt when body posture does not change. However, during prolonged measurements of Vt, the accuracy of RIP diminishes. Whyte et al 9 studied eight healthy individuals during sleep and found a poor correlation between Vt measured by RIP and by a pneumotachograph (mean R to 0.36). The inaccuracies were attributed to changes in posture during sleep and slippage of the RIP belts. These same factors may account for discrepancies between flowmeter and RIP measurements of Vt during cycling (R ) and treadmill exercise 688 Clinical Investigations in Critical Care

6 Figure 4. The differences between Vtmag and Vtspiro are plotted for each breath at rest (n 1,111; top) and with exercise (n 1,163; bottom). The mean difference between Vtmag and Vtspiro is depicted by the solid line, and the 95% confidence intervals ( 2 SD) are depicted by the dashed lines. (R ). 10 The above-mentioned observations suggest that a two-degrees-of-freedom model can be used for semiquantitative purposes during prolonged monitoring in a given body position but not to quantitate Vt. Such semiquantitative measures have been used to identify hypopneas and apneas during sleep. 11 Volume measurements using the standard twodegrees-of-freedom model can be significantly improved by adding an axial measure of chest wall motion (Xi). Changes in the axial distance reflect the changes in chest wall shape that occur with changes in posture. This axial distance is measured as the distance between the xiphoid and the pubic symphy- CHEST / 122 / 2/ AUGUST,

7 Table 4 Mean Difference and Limits of Agreement Variables Mean Value (Magnetometer) SD 2 Mean Difference Limits of Agreement Exercise Vt, ml 1, Resting Vt, ml 1, Ti, s Te, s sis or between the xiphoid and the umbilicus. 2,7 When the three-degrees-of-freedom model is tested in a laboratory setting, values of Vt obtained during maneuvers associated with changes in spinal attitude (eg, standing-sitting, lifting, and walking) were generally within 3% of spirometric values with R 2 values Similarly, the mean values of Timag were within 7% of Tispiro with R 2 values The feasibility of using the three-degrees-of-freedom model for more prolonged periods (1 to 2 h) outside the laboratory has been demonstrated in a field study of nine subjects engaged in auto-body work such as sanding and spray painting. Group mean values of minute ventilation (V e) were within 20% of that measured by a pneumotachometer during these tasks. 5 However, the equipment used to measure and record body surface displacement was cumbersome, bulky, loaded on to large carts, and required wall outlets for power. This technology was not well suited for field use and likely impeded subject activity. The present study was designed to test the accuracy of a magnetometer device that had been reengineered for field use. In comparison with the previous magnetometer device, the circuitry for the transmitter and receiver coils had been miniaturized, the number of coils reduced to four, a flowmeter incorporated into the device, data stored in compact flash memory, and the power requirements reduced. Our results show that this ventilation monitor could measure Vt, Ti, and Te with subjects seated, standing, and walking at a brisk pace with reasonable accuracy. The best correlations between Vtspiro and Vtmag were obtained when using a calibration maneuver that was position specific, ie, calibration coefficients determined in the seated position were applied to data obtained while the subject was seated. However, the correlation between these two variables remained reasonably good when applying the seated calibration coefficients to data obtained with subjects who were standing (R ). These better than expected correlations may be attributed to the large range of spinal attitude that is included in the calibration maneuver when subjects are instructed to flex and extend the spine. This type of calibration maneuver may be more appealing than a position specific calibration maneuver when changes in body position are expected to occur while monitoring breathing. The accuracy of this device was evaluated over a wide range of Vts as subjects changed their posture or exercised. The greatest inaccuracies in measurements of Vt were noted with Vts 2,500 ml (Fig 4, top and bottom). The vast majority of data points outside the 95% confidence intervals were at these higher Vts, whereas very few points were outside the 95% confidence intervals over the lower range of Vt in either the rest or exercise conditions. The errors in measuring Vt at the higher Vts broadened the limits of agreement at the lower Vts. The inaccuracies at the higher range of Vt may be caused by differences in breathing pattern between the calibration and testing periods. The range of Vt during the calibration maneuver was smaller than the range of Vt encountered during the testing periods (either at rest or with exercise). Second, the rib cage and abdomen may have distorted more at the greatest Vt. This would produce more than three degrees of freedom of chest wall motion and may reduce accuracy at the highest Vt. Chest wall distortion during exercise also may explain, in part, why the exercise data were not as accurate as the resting data. Calibration maneuvers performed with individuals specifically instructed to breathe with large Vts may enhance the accuracy of this device over this range. Potential applications for this device would include measuring V e in the workplace or at home. Currently, there are no commercially available portable devices that can be used to assess axial changes in the chest wall and to noninvasively measure Vt. Given that most of the inaccuracies of the ventilation monitor were with Vts 2.5 L, and that Vts 2.5 L are unlikely to be encountered at rest or during light activity, this device may be acceptable for use by an epidemiologist or industrial hygienist interested in using V e as a means of assessing exposures to air toxics. Such direct measures of V e and Vt would be more accurate than those derived from measurements of heart rate. 5 Another potential application for this device is for measurements of breathing patterns during sleep. Equipment cur- 690 Clinical Investigations in Critical Care

8 rently used in sleep laboratories only assesses two degrees of freedom of chest wall motion. Changes in posture during sleep, then, will lead to inaccurate measurements of Vt. The degree of accuracy that we demonstrated with the ventilation monitor should be acceptable for detecting apneas and hypopneas during sleep. In some individuals, with the most accurate calibration maneuvers, this device may provide an alternative technique to measure Vt during exercise that avoids the use of bulky equipment or mouthpieces. In conclusion, an unencumbering portable magnetometer system can be used to measure Vt, Ti, and Te over a sixfold range of Vts during breathing in seated, standing, and exercising subjects. We speculate that such a device will facilitate measurements of breathing pattern in environments outside the laboratory. References 1 Konno K, Mead J. Measurement of the separate volume changes of rib cage and abdomen during breathing. J Appl Physiol 1967; 22: Smith JC, Mead J. Three degree of freedom description of movement of the human chest wall. J Appl Physiol 1986; 60: McCool FD, Kelly KB, Loring SH, et al. Estimates of ventilation from body surface measurements in unrestrained subjects. J Appl Physiol 1986; 61: Paek D, McCool FD. Breathing patterns during varied activities. J Appl Physiol 1992; 73: McCool FD, Paek D. Measurements of ventilation in freely ranging subjects. Res Rep Health Eff Inst 1993; 59: Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurements. Lancet 1986; 2: Paek D, Kelly KB, McCool FD. Postural effects on measurements of tidal volume from body surface displacements. J Appl Physiol 1990; 68: Zimmerman PV, Connellan SL, Middleton HC, et al. Postural changes in rib cage and abdomen volume-motion coefficients and their effect on calibration of a respiratory inductance plethysmograph. Am Rev Respir Dis 1983; 127: Whyte KF, Gugger M, Gould GA, et al. Accuracy of respiratory inductive plethysmograph in measuring tidal volume during sleep. J Appl Physiol 1991; 71: Caretti DM, Pullen PV, Premo LA, et al. Reliability of respiratory inductive plethysmography for measuring tidal volume during exercise. Am Ind Hyg Assoc J 1994; 55: Cantineau JP, Escourrou P, Sartene R, et al. Accuracy of respiratory inductive plethysmography during wakefulness and sleep in patients with obstructive sleep apnea. Chest 1992; 102: CHEST / 122 / 2/ AUGUST,