Kohler disease

Objectives: To evaluate the accuracy of a portable respiratory inductive plethysmograph that allows the monitoring of ventilation without airway instrumentation during exercise in unrestrained subjects.

Design: Validation of a novel technique by comparison to a reference standard.

Participants: Thirty-one subjects, including 20 healthy volunteers, 6 patients with COPD, and 5 patients with congestive heart failure.

Interventions: Participants performed progressive treadmill exercise to exhaustion. Ventilation was monitored by a novel battery-powered, miniaturized, and calibrated respiratory inductive plethysmograph. Inductance sensors encircling the rib cage and abdomen were built into an elastic body garment. A pneumotachograph attached to a mouthpiece served as the reference method.

Measurements and results: Breath-by-breath comparisons between the inductance plethysmograph and pneumotachograph over the course of progressive exercise to exhaustion revealed no significant bias of respiratory cycle time, tidal volume (VT), and minute ventilation. The corresponding limits of agreement (bias [+ or -] 2 SDs) were [+ or -] 6%, [+ or -]17%, and [+ or -] 17%, respectively, for 2,480 breaths. Comparisons of mean values averaged over 20 breaths revealed improved limits of agreement of [+ or -] 1% for cycle time, and [+ or -] 7% for tidal volume and minute ventilation, respectively, for 124 comparisons. Agreement between methods was similar for patients and healthy subjects. Among the patients, maximal minute ventilation was lower, and breathing was more rapid and shallow than in healthy subjects. Obstructive lung disease was associated with a shorter duty cycle than heart failure.

Conclusions: The portable respiratory inductive plethysmograph accurately estimates ventilation during treadmill exercise, and identifies differences in breathing patterns among patients with pulmonary or cardiac diseases and healthy subjects. This unobtrusive monitoring technique is promising for application in ambulatory patients.

Key words: breathing pattern; exercise; inductive plethysmography; noninvasive physiologic monitoring; pulmonary ventilation

Abbreviations: CHF = congestive heart failure; RIP = respiratory inductive plethysmography TI/TT = duty cycle TT = total duration of respiratory cycle; VI = inspired minute ventilation; V[o.sub.2]max = maximal oxygen uptake; VT = tidal volume

**********

Spirometric assessment of lung function and evaluation of the ventilatory response to exercise by standard techniques requires patient cooperation and airway instrumentation. Since these measurements alter the breathing pattern and ventilation, (1,2) and are usually performed in a laboratory, they may not reflect ventilation during usual daily physical activities of patients at their home or workplace. Portable equipment for unobtrusive and cooperation-independent monitoring of ventilation in ambulatory patients would therefore be desirable. Chest surface sensors of an inductive plethysmograph or magnetometer have been used for noninvasive ventilatory monitoring during sleep and for physiologic research, but the application of these techniques in an ambulatory setting has been hampered by the lack of portable equipment and by the tendency of conventional sensors to dislocate with body motion. (3-6)

Recently, a portable respiratory inductive plethysmography (RIP) system has become commercially available. It comprises a novel patient interface consisting of a snugly fitting, elastic garment incorporating rib cage and abdominal inductance sensors, thereby minimizing the propensity for sensor dislocation. The corresponding battery-powered digital signal-processing unit has the weight and size of a palmtop computer. Since this equipment seems promising for application in outpatients under various conditions at rest, and during exercise, the purpose of this study was to evaluate its accuracy for measuring tidal volume (VT) and respiratory timing during exercise. We have previously demonstrated the accurate measurement of ventilation by RIP during maximal bicycle exercise in a small number of healthy subjects. (1) To extend the validation of RIP to the measurement of ventilation during a form of physical activity that is more representative of the daily activities of many subjects, the current protocol included treadmill exercise. Furthermore, we studied a larger number of healthy volunteers as well as patients with cardiac and pulmonary diseases to evaluate the effect of the disease and associated abnormalities of chest wall motion and breathing pattern (6-8) on the accuracy of ventilatory monitoring by RIP. We hypothesized that the measurement of ventilation by RIP in healthy subjects and patients during exercise is valid within the published standards.

Lung volume changes are commonly estimated by body surface sensors of magnetometers or RIP sensors placed at the rib cage and abdomen based on the classic two-degrees-of-freedom model of chest wall excursions described by Konno and Mead. (9) Smith and Mead, (10) and McCool and coworkers (11) proposed a three-degrees-of-freedom model that included an additional term reflecting axial chest wall motion related to flexion and rotation of the vertebral spine and pelvis. To investigate the potential effect of such movements on the accuracy of RIP (with rib cage and abdominal sensors only), we compared measurements among volunteers carrying a custom-fitted and loaded backpack incorporating a metal frame that stabilized their back, and volunteers with unrestricted spinal motion. We hypothesized that restricting spinal motion by the backpack would not significantly improve the agreement between RIP and flowmeter-derived estimates of ventilation.

MATERIALS AND METHODS

Subjects and Patients

A total of 31 individuals, among them 20 healthy volunteers, 5 patients with severe congestive heart failure (CHF), and 6 patients with COPD (Table 1) gave informed consent to participate in the study, which was approved by the hospital ethics committee. The healthy volunteers were recruited from among hospital staff. Any medical condition requiring treatment at the time of the study was an exclusion criterion. The patients were recruited from among successive patients who had been referred for cardiopulmonary exercise testing to evaluate physical performance and dyspnea during exercise. Of the five patients with heart failure, three were studied during regular follow-up and two were evaluated for possible cardiac transplantation. Three had severe coronary artery disease from dilated cardiomyopathy. Four of the six COPD patients were candidates for lung volume reduction surgery, and two were referred for evaluation of progressive symptoms.

Measurements

RIP sensors consisted of two shielded electrical conductors that were sewn in a sinusoidal array into a sleeveless, snugly fitting elastic shirt so that they encircled the rib cage and abdomen at the level of the nipples, and the umbilicus (LifeShirt; VivoMetrics; Ventura, CA) [Fig 1]. Sensors were placed by one of the investigators and were connected to a portable, battery-powered device incorporating RIP, pulse oximetry, ECG, three accelerometers for the measurement of body movements and position in three axes, and a signal-processing and recording unit with the capacity for continuous data collection over [greater than or equal to] 24 h. The device is the size of a palmtop computer (dimensions, 17.3 x 7.9 x 3.8 cm; weight, 382 g) and was attached to a belt.

[FIGURE 1 OMITTED]

A spiroergometry unit (Vmax; SensorMedics; Yorba Linda, CA) served as the reference standard. Calibration and breath-by-breath measurements of ventilation and oxygen uptake were performed as previously described. (12) During exercise tests, subjects breathed through the flowmeter attached to a mouthpiece while the nose was clipped.

Protocol

Ventilation was simultaneously monitored by RIP and the flowmeter during progressive treadmill exercise to exhaustion. The initial speed was 1.5 miles per hour, and increased approximately every minute by 0.5 to 1 mile per hour. The treadmill was inclined at 12[degrees] in healthy subjects, and between 0[degrees] and 3.5[degrees] in patients to achieve an exercise duration of at least 5 min. Subjects and patients were encouraged to exercise to their exhaustion, but no other instructions were given. Tests were terminated according to standard criteria, (13) and when the subject or patient could no longer keep up with the speed and wished to stop. Ten of the 20 healthy subjects were carrying a climber's backpack loaded to a weight of 11 kg. It had a built-in metallic frame that stabilized their back, and was closely attached to the body by the shoulder harness and a belt around the abdomen, located approximately at the iliac crest.

Data and Statistical Analysis

At the end of tests, the data were downloaded from the RIP recorder to a computer and were analyzed by dedicated software (VivoLogic; VivoMetrics). It allowed the review of raw and processed data on a video screen, automatic calibration, and breath-by-breath calculation of breathing pattern variables and heart rate as previously described. (1,6) Relative gains in rib cage and abdominal RIP signals were determined at the beginning of the tests during 5 min of regular breathing performed in a standing position by the qualitative diagnostic calibration method. (14) The RIP sum signal was calibrated in absolute units (liters) by comparison to the integrated output of the flowmeter over the duration of 20 breaths. (1) Inspiratory VT, total duration of respiratory cycle (TT), and inspiratory minute ventilation (VI), which is the product of VT and breath rate, were measured breath by breath as described previously. Rib cage abdominal motion was assessed by computing the rib cage contribution to VT, and the phase shift between rib cage and abdominal excursions, as described recently. (15) The heart rate was determined by the ECG incorporated in the portable RIP recorder. The spiroergometry unit measured respiratory variables breath by breath, and these data were matched with corresponding values obtained with the RIP device by means of a common time stamp on the ECG signals of both systems that synchronized the recorders.

To compare equal numbers of breaths from each individual despite the variation in exercise duration, four periods of 20 breaths from low, intermediate, submaximal, and maximal exercise intensity, respectively, were analyzed (ie, a total of 80 breaths per individual). Data collected after the initiation of walking when the breathing pattern had stabilized corresponded to low intensity. Intermediate intensity was defined as a period within the first half of the duration of exercise after low intensity. Submaximal data were collected within the second half of the duration of exercise, before maximal exercise. Maximal values of physiologic variables were defined as their mean over the final 20 breaths during maximal exercise. The predicted maximal heart rate was calculated as 220- age, and the predicted maximal oxygen uptake (V[o.sub.2]max) was computed according to Wassermann et al. (16)

The data are summarized as the mean [+ or -] SD. Agreement between methods was evaluated by computing the mean difference (bias) and limits of agreement ([+ or -] 2 SDs of the bias). (17) The mean disagreement between methods was computed as the absolute difference in percent of the mean value by the two methods. Comparisons between groups were performed by analysis of variance followed by the Newman-Keuls test, where appropriate. A probability of p < 0.01 was considered to be significant.

With a sample size of 2,480 comparisons (for breath-by-breath data), and 124 comparisons (for mean data averaged over the duration of 20 breaths), the study was powered with > 0.95 to detect a [greater than or equal to] 2% bias between methods in VT and VEi, and a bias of [greater than or equal to] 1% in TT ([alpha] 0.05). To evaluate whether a nonsignificant result of a comparison between the two methods can be considered as a true absence of difference (ie, equivalence), we employed the confidence interval approach, as described by Jones et al. (18) According to this method, equivalence is assumed if the confidence interval of an observed difference lies entirely within a specified equivalence range, which was defined as [+ or -] 2% for VT and VI, and as [+ or -] 1% for TT.

RESULTS

A satisfactory snug fit of the available standard sizes of RIP garments was achieved in all subjects and patients, and RIP recordings were successfully obtained in all of them. Figure 2 shows recordings in a healthy female subject, and a male patient with COPD. Table 2 summarizes V[o.sub.2] max, heart rate, and breathing pattern variables over the final 20 breaths during maximal exercise. The V[o.sub.2] max was above the predicted value in healthy subjects, and their maximal heart rate was close to the predicted value achieved at the end of the exercise tests. In contrast, V[o.sub.2] max was significantly reduced in patients, in particular in those with severe CHF, and they did not reach their predicted maximal heart rate. All patients with CHF were on [beta]-blocking drugs. The highest VI and breath rate were observed in healthy subjects carrying a backpack. They achieved their higher VI compared to healthy subjects not carrying a backpack through a higher breath rate, while VT was lower. Patients with COPD had a higher breath rate/VT ratio (ie, a more rapid shallow breathing pattern) and a shorter duty cycle (TI/TT) than healthy subjects and patients with CHF (Table 2). In addition, patients with COPD had a greater rib cage contribution to VT and a greater asynchrony of rib cage vs abdominal motion (quantified by the phase angle) (6,15,19) than patients with CHF.

[FIGURE 2 OMITTED]

Table 3 summarizes the agreement between RIP and flowmeter-derived breath-by-breath VI, VT, and TT for 80 breaths in each individual, including 20 breaths each during low, intermediate, submaximal, and maximal treadmill exercise. There was no statistically significant bias in values found by the two methods in any of the groups of subjects and patients, and the 95% confidence intervals of differences among the biases in these groups was < [+ or -] 2% for VT and VI, and < [+ or -] 1% for TT, suggesting equivalent agreement. (18) In other words, the presence of disease did not affect the accuracy of the RIP. The ranges of the limits of agreement for breath-by-breath comparisons in the different groups were [+ or -] 5 to [+ or -] 8% for TT, and [+ or -] 16 to [+ or -] 18% for VT and VEi.

Comparisons of mean values from the two methods averaged over the duration of 20 breaths per individual during low, intermediate, submaximal, and maximal exercise, respectively, revealed close agreement with all paired values of TT falling within 5% of each other, and all values of VT and VI falling within 10% of each other (Fig 3). There was no significant bias, and the limits of agreement were [+ or -] 1% for TT, and [+ or -] 8% for VT and VI. Moreover, in each group of healthy subjects and patients, there was a close correlation among the values for TT, VT, and VI by the methods, with [r.sup.2] exceeding 0.97 (p < 0.001). Since the mean data over the final 30 to 60 s of maximal exercise are commonly analyzed for the assessment of maximal exercise performance, agreement between measurements made with RIP and the flowmeter for the mean TT, VT, and VI over the final 20 breaths at maximal exercise (corresponding to approximately 30 s and to the data of Table 2) are presented separately in Table 4.

[FIGURE 3 OMITTED]

DISCUSSION

The novel features of the respiratory monitoring system evaluated in the current study are the portability related to a miniaturized, battery-powered signal-processing and signal-recording unit, the patient interface, consisting of inductance sensors integrated in a snugly fitting garment to minimize dislocation, and software for semiautomatic breathing pattern analysis. Our data demonstrate a high accuracy of portable RIP-derived estimates of TT, VT, and ventilation during maximal treadmill exercise in healthy subjects and in patients with severe cardiac and pulmonary diseases. Moreover, the analysis of RIP revealed characteristics of the breathing pattern that differed among healthy subjects and patients with cardiac or pulmonary disease. Our observations suggest that the portable RIP system is suitable for monitoring breathing patterns and ventilation in unrestrained subjects and patients during exercise.

The V[O.sub.2] max and heart rate over the last 9.0 breaths during maximal exercise indicated a high intensity of exercise performed by the healthy subjects (Table 2). In patients with severe CHF, heart rate might not have reached the predicted maximal value because of the treatment with [beta]-blocking agents or because of premature exercise termination due to leg weakness or insufficient effort. In patients with COPD, exercise was most likely limited by their inability to increase ventilation. Consistent with the exhausting ventilatory effort they performed, their breathing pattern was more rapid and shallow than that of patients with CHF or healthy subjects, as quantified by a high breath rate/VT ratio (Table 2). Further characteristics that differentiated patients with COPD from those with CHF were a short TI/TT, a high rib cage contribution to VT, and asynchronous rib cage-abdominal motion (ie, a greater phase angle) [Table 2]. These breathing patterns were consistent with our previous observations in patients with COPD at rest. (6)

With no significant bias for breath-by-breath estimates of TT, VT, and VI, and limits of agreement of [less than or equal to] 8% for TT and [less than or equal to] 18% for both VT and VI (Table 3), the accuracy of portable RIP in the current investigation compares favorably to results achieved in previous studies using stationary equipment in resting patients with chronic obstructive lung disease undergoing lung volume reduction surgery, (6,15) in sleep apnea patients during sleep studies, (4,5) and in healthy volunteers during bicycle exercise (1) and treadmill exercise. (20) In an earlier study, (1) we had to tape the RIP sensors encircling the rib cage and abdomen to the skin to avoid displacement during exercise. Major excursions in the abdominal RIP signal related to leg motion with pedaling were noted (Fig 2 in our previous publication). (1) This required adjustments of the criterion for automatic, computer-assisted breath detection with progressive exercise, In the current study, no special fixation of RIP sensors contained within the garment was necessary. Walking at a brisk pace on the treadmill was also associated with artifacts in RIP raw waveforms (Fig 2). However, this did not require adjustment of the minimal volume change accepted as a breath by the computer algorithm from the default value of 25% of VT during the initial qualitative diagnostic calibration. (1,14) This default value was preset in the software (VivoLogic), which employed similar concepts for automatic breath detection as we have previously described. (1) The robust computer-assisted breath detection achieved in the current study during treadmill exercise may have been related to a more favorable signal/noise ratio compared to bicycle exercise, and to the fact that the software (VivoLogic) applies digital filtering to RIP signals before performing breath detection and computation of breathing pattern variables (see the VivoLogic software user manual). It is also conceivable that the leg and body movements between bicycle and treadmill exercise might have been associated with different amounts of movement artifacts on respiratory waveforms.

According to recommended standards, the assessment of exercise performance is not based on breath-by breath values but on data averaged over > 30 to 60 s. (13) Therefore, the comparison of mean RIP and flowmeter-derived values averaged over periods of 20 breaths (corresponding to approximately 30 s at maximal exercise) were also performed. This analysis revealed limits of agreement of [+ or -] 1% for TT, and [+ or -] 8% for VT and VI (Fig 3). Thus, 95% of differences were expected to fall within these ranges. This indicates, for example, that a reduction in the ventilatory reserve measured by a flowmeter during maximal exercise by > 8% below the expected value will be detected by RIP in 97.5% of cases. Taking into account a tolerated inaccuracy of flowmeters within [+ or -] 5% (according to American Thoracic Society standards), (21) and repeatability coefficients of 25 to 28% for maximal minute ventilation measured by a flowmeter during repeated bicycle exercise tests in the same individuals, 1 week apart, (12) the accuracy and precision of RIP achieved in the current study is excellent. Whether similar results can be obtained in other, less controlled settings, over longer time periods, and in more obese patients remains to be confirmed.

Based on measurements in healthy volunteers performing specific respiratory maneuvers and activities that induced rotation or flexion/extension of the vertebral spine and pelvis, Smith and Mead, (10) and McCool and coworkers (11,22) suggested that the estimation of lung volume changes based on data from body surface sensors could be improved by monitoring axial displacements by a magnetometer attached to the skin at the sternal and umbilical level as a third degree of freedom of chest wall motion, in addition to anteroposterior dimensions (measured by magnetometers) or cross-sections of the rib cage and abdomen (measured by RIP). Employing a portable ventilation monitor incorporating two pairs of magnetometers arranged in an elegant array so that their signals reflected anteroposterior rib cage and abdominal dimensions as well as the sternal-umbilical distance, McCool and coworkers (22) achieved a satisfactory agreement of VT estimates compared to flowmeter measurements during submaximal bicycle exercise in nine healthy volunteers. At a mean VT of 1.74 L, the bias and the limits of agreement were 0.18 and [+ or -] 0.69 L, respectively, corresponding to a mean relative disagreement of 14%. (22) In the current study involving maximal exercise, the bias and limits of agreement computed for all participants, including healthy subjects and patients, compare favorably with the cited results (the bias and the limits of agreement of VT were -0.01 [+ or -] 0.32 L, the mean disagreement was 7% at a mean VT of 1.93 L) [Table 3]. The accuracy of RIP did not differ among patients and healthy subjects without a backpack, and healthy subjects carrying a backpack that restricted the range of spinal motion by the metal frame and weight (Table 3). Therefore, the degrees of freedom of chest wall motion in addition to those monitored by the two RIP sensors, presumably induced by rigorous muscle contractions during maximal exercise, an elevated respiratory impedance in patients, or spinal flexion/extension or rotation, did not significantly affect lung volume estimates by RIP during maximal treadmill exercise.

In conclusion, we have demonstrated accurate, cooperation-independent measurement of ventilation by a novel, portable monitoring device during progressive treadmill exercise to exhaustion in healthy subjects and patients with cardiac and pulmonary diseases. The RIP monitor was able to identify differences in breathing patterns between healthy subjects and patients with cardiac or pulmonary disease. Therefore, portable RIP is a promising tool for unobtrusive respiratory monitoring in unrestrained subjects and patients for investigation of ventilatory adaptation to exercise and disease.

REFERENCES

(1) Bloch KE, Barandun J, Sackner MA. Effect of mouthpiece breathing on cardiorespiratory response to intense exercise. Am J Respir Crit Care Med 1995; 15:1087-1092

(2) Perez W, Tobin MJ. Separation of factors responsible for changes in breathing pattern induced by instrumentation. J Appl Physiol 1985; 59:1515-1520

(3) Sackner MA, Krieger BP. Noninvasive respiratory monitoring. In: Scharf SM, Cassidy SS, eds. Heart-lung interactions in health and disease. New York, NY: Marcel Dekker Inc, 1989; 663-805

(4) Bloch KE, Li Y, Sackner MA et al. Breathing pattern during sleep disruptive snoring. Eur Respir J 1997; 10:576-586

(5) Kaplan V, Zhang JN, Russi EW, et al. Detection of inspiratory flow limitation during sleep by computer assisted respiratory inductive plethysmography. Eur Respir J 2000; 15:570-578

(6) Bloch KE, Li Y, Zhang Jet al. Effect of surgical lung volume reduction on breathing patterns in severe pulmonary emphysema. Am J Respir Crit Care Med 1997; 156:553-560

(7) Hoover CF. Diagnostic significance of inspiratory movements of costal margins. Am J Med Sci 1920; 159:633-646

(8) Tobin MJ, Chadha TS, Jenouri G, et al. Breathing patterns: 2. Diseased subjects. Chest 1983; 84:286-294

(9) Konno K, Mead J. Measurement of separate volume changes of ribcage and abdomen during breathing. J Appl Physiol 1966; 22:407-422

(10) Smith JC, Mead J. Three degree of freedom description of movement of the human chest wall. J Appl Physiol 1986; 60:928-934

(11) McCool FD, Kelly KB, Loring SH, et al. Estimates of ventilation from body surface measurements in unrestrained subjects. J Appl Physiol 1986; 6:1114-1119

(12) Bingisser R, Kaplan V, Scherer T, et al. Effect of training on repeatability of cardiopulmonary exercise performance in normal men and women. Med Sci Sports Exerc 1997; 29: 1499-1504

(13) American Thoracic Society, American College of Chest Physicians. ATS/ACCP statement on cardiopulmonary exercise testing. Am J Respir Crit Care Med 2003; 167:211-277

(14) Sackner MA, Watson H, Belsito AS, et al. Calibration of respiratory inductive plethysmograph during natural breathing. J Appl Physiol 1989; 66:410-420

(15) Brack T, Senn O, Russi EW, et al. Transtracheal high flow insufflation supports spontaneous respiration in chronic respiratory failure. Chest 2005; 127:98-104

(16) Wasserman K, Hansen JE, Sue DY, et al. Principles of exercise testing and interpretation. 2nd ed. Philadelphia, PA: Lea & Febiger, 1994

(17) Bland MJ, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-310

(18) Jones B, Jarvis P, Lewis JA, et al. Trials to assess equivalence: the importance of rigorous methods. BMJ 1996; 313:36-39

(19) Agostoni E, Magnoni P. Deformations of the chest wall during breathing efforts. J Appl Physiol 1966; 21:1827-1832

(20) Sackner JD, Nixon AJ, Davis B, et al. Non-invasive measurement of ventilation during exercise using a respiratory inductive plethysmograph. I. Am Rev Respir Dis 1980; 122:867-871

(21) American Thoracic Society. Standardization of spirometry, 1994 update. Am J Respir Crit Care Med 1995; 152:1107-1136

(22) McCool FD, Wang J, Ebi KL. Tidal volume and respiratory timing derived from a portable ventilation monitor. Chest 2002; 122:684-691

* From the Pulmonary Division, Department of Internal Medicine, University Hospital Zurich, Zurich Switzerland.

Supported by grants from the Zurich Lung League and the Hartmann-Muller Stiftung Zurich. The RIP monitor (LifeShirt) was provided by VivoMetrics, Ventura, CA.

Manuscript received October 16, 2004; revision accepted February 8, 2005.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml).

Correspondence: Konrad E. Bloch, MD, FCCP, Pulmonary Division, Department of Internal Medicine, University Hospital of Zurich, Ramistrasse 100, CH-8091 Zurich, Switzerland; e-mail: pneubloc@usz.unizh.ch

COPYRIGHT 2005 American College of Chest Physicians
COPYRIGHT 2005 Gale Group