INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Inspiratory muscle function during exercise has been extensively studied in normal subjects and patients with cardiorespiratory diseases. The tension-time index of the diaphragm, introduced by Bellemare and Grassino (1, 2), has been assessed during exercise in normal subjects with or without supplemental oxygen administration (3) and with or without restrictive loads (4), as well as in patients with chronic heart failure (5). During exercise, normal subjects and patients with chronic obstructive pulmonary disease (COPD) preferentially increase rib cage muscle pressure contribution to meet the ventilatory demands (6). In this case the global activity of the inspiratory muscles can be assessed by a tension-time index of inspiratory muscles (TTI), given by the equation /Pes_max × TI/Ttot, where is mean inspiratory esophageal pressure, _max is maximal static Pes, TI is inspiratory time, and Ttot is total duration of the breathing cycle. Such a tension-time index, based on esophageal pressure, has also been used in normal subjects and patients with COPD (7). These two indices have been used to assess not only the risk of inspiratory muscle fatigue (3), but also the inspiratory muscle effort (5, 7, 8), and they may therefore provide useful information on mechanisms involved in exertional dyspnea (8, 9). Since such studies are invasive because they require the use of a gastric or esophageal balloon, or both, they cannot be used in a large number of subjects and thus are not used routinely in clinical assessment. Our group has validated a noninvasive index of inspiratory muscle activity (TTMUS) based on measurement of mouth pressure in resting healthy subjects and patients with COPD (10). In this index, the mean inspiratory pressure (PI) is derived from measurement of the mouth occlusion pressure (P_0.1) according to an equation proposed by Gaultier and coworkers: PI = 5 P_0.1 × TI (11). In this equation the rise in pressure during inspiration is approximated by a single power function of time, assuming a linearly increasing pressure profile. In resting conditions this method has been shown to be useful in assessing inspiratory muscle function in healthy children and adults and in patients with cystic fibrosis or adults with COPD (12, 13). However, during exercise in healthy subjects and patients with COPD, the pressure-time course has been shown to exhibit an upward convexity (14, 15). When the inspiration driving pressure increases nonlinearly with respect to time, the PI is overestimated by extrapolating the P_0.1 over the entire TI. The use of this initial method (10) during exercise may thus lead to a marked overestimation of the actual mean pressure developed by the inspiratory muscles. The continuing need for a noninvasive approach thus led us to modify this method. We previously reported a significant correlation at rest between P_0.1 and in normal subjects and patients with COPD (10). Hussain and coworkers (16) showed in healthy subjects during exercise that P_0.1 accurately reflects the changes in activity of the major inspiratory muscles. Therefore, in the present study we investigated whether a noninvasive tension-time index, in which changes in P_0.1 alone are used to assess changes in , would be reliable to assess the overall inspiratory muscle activity during exercise. This index, labeled TT_0.1, is given by P_0.1/ PI_max × TI/Ttot, where PI_max is the maximal inspiratory pressure. The aim of this study was to assess the validity of TT_0.1 during exercise in normal subjects and patients with COPD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

We first studied nine male patients with COPD and seven male normal volunteers whose anthropometric characteristics and lung function data are given in Table 1. The diagnosis of COPD was made according to the American Thoracic Society guidelines (17). The patients were free from other cardiopulmonary disease, and at the time of the study all were in a clinically stable state without exacerbation for at least 1 mo. Predictability and reproducibility of our method were evaluated in another group of five normal males having characteristics similar to those described in Table 1. The study was approved by the local ethics committee, and each subject gave informed consent.

Measurements

In the patients with COPD the TLC and the FRC were measured with the helium dilution method (Pulmonet III; SensorMedics, Anaheim, CA). FEV₁ and FVC were measured in all subjects according to standard spirographic technique and procedures (18). The predicted values were those of the European Community for Steel and Coal (18).

Subjects were studied while breathing through a one-way low-resistance valve (0.9 cm H₂O · L¹ · s; dead space: 50 ml; Warren E. Collins Inc., Braintree, MA). Flow was measured with a pneumotachograph (No. 3; Fleisch, Lausanne, Switzerland) placed on the inspiratory line of the valve connected to a differential pressure transducer (Model MP45, ± 2 cm H₂O; Validyne Corp., Northridge, CA). Tidal volume (VT) was obtained by integration of the flow signal. A noncommercial silent electromagnetic valve, made in the laboratory of Dr. J. Milic-Emili (Meakins-Christie Laboratories, Montreal, QC, Canada), was used to perform mouth occlusions. The valve was placed on the inspiratory line of the one-way valve, and was closed during expiration and opened automatically ~ 150 ms after the onset of the occluded inspiration. The pressure during occluded inspirations was measured with a MP45 (± 35 cm H₂O) Validyne differential pressure transducer and a model CD 15 carrier demodulator (Validyne Corp., Northridge, CA).

Esophageal pressure was measured with an 8-cm-long, thin-walled latex balloon attached to a polyethylene catheter (1.4 mm interior diameter, Marquat, Boissy-St-Leger, France) positioned in the lower third of the esophagus (19). The catheter was connected to one side of a MP45 (± 150 cm H₂O) Validyne differential pressure transducer, with the other side open to the atmosphere. Gastric pressure (Pgas) was also measured in the patients with COPD using an 8-cm-long thin-walled latex balloon positioned in the stomach. With this equipment, flow and pressure measurements were not affected by phase shift or alteration in amplitude up to 20 Hz.

The PI_max and Pes_max pressure were measured at FRC in the sitting position with a MP45 (± 300 cm H₂O) Validyne differential pressure transducer, using the technique of Black and Hyatt (20). All normal subjects but one and all patients with COPD had no previous experience of these maneuvers. Therefore, great care was taken to fully explain the procedures. This was facilitated by the use of an oscilloscope (Gould Inc., Cleveland, OH) which provided a visual feedback to the subjects during the explanations and during the maneuvers. The subjects were asked to perform a maximal inspiratory effort against an occluded airway and to maintain the maximal pressure for at least 1 s. Repeated measurements were made until 3 technically satisfactory and reproducible measurements were obtained (variation < 10%). The reported data represent the best values.

The exercise test was performed on a bicycle ergometer (Ergometer 990; Bodyguard Jones AS, Sandnes, Norway). During the test, the subjects wore a nose clip and breathed through the valve, which was connected to a breath-by-breath automated exercise metabolic system by the expiratory circuit (CPX; Medical Graphics Corp., St. Paul, MN). Oxygen uptake (O₂), carbon dioxide output, and respiratory ratio were continuously measured using the CPX. Before each test, the gas analyzers were calibrated with two gas mixtures of known oxygen and carbon dioxide concentration. The data were averaged during the last 20 s of each load over an integral number of breaths. The electrocardiogram was monitored with a cardioscope (Personal 120; Esaote-Biomedica, Florence, Italy).

Protocol

After the lung function measurements, the normal subjects and patients with COPD underwent tests with pressure measurements. After insertion of gastric or esophageal balloons, or both they were studied at rest seated on a chair. After they had been accustomed to the experimental equipment (mouthpiece and nose clips), and the respiratory ratio was stable, ventilatory and pressure parameters were recorded for 5 min. At least 10 occlusions were then performed in each subject, at the rate of 2 to 3 per min. During another 5-min period the measurements were repeated. The subjects performed the PI_max and Pes_max maneuvers after a 5-min period.

The subjects next performed an incremental exercise test on a cycle ergometer. Measurements at rest were obtained after a 5-min resting period on the cycle, after a stable respiratory ratio was attained. This was followed by a 3-min 20-W warm-up period for the patients with COPD and 30 W for the control subjects. The workload was then increased in steps of 10 W every 90 s for the patients with COPD and 30 W every 90 s for the control subjects, until exhaustion. All subjects were encouraged to exercise until they felt unable to continue. At each level of exercise, a 90-s period was needed to obtain the measurements of respiratory gas exchange, breathing pattern, mouth and esophageal pressure, and P_0.1.

Analysis

All signals were displayed on a Gould ES1000 recorder (Gould Inc., Cleveland, OH) with a paper speed of 10 to 25 mm · s¹ for measurements of ventilatory and pressure parameters, and of 100 mm · s¹ for measurements of mouth occlusions.

At rest, VT, respiratory frequency (f), minute ventilation (E), TI, Ttot TI/Ttot, and were determined from an average of 10 respiratory cycles before and after the mouth occlusion measurements. TI and Ttot were obtained from the flow signal. The P_0.1 value at rest corresponds to the average of all measurements.

During exercise, the ventilatory and esophageal pressure parameters were determined from an average of 10 respiratory cycles during the last minute of each workload. The exercise P_0.1 represents the average of 3 to 5 measurements made during this period.

The was obtained by averaging the pressure at intervals of 200 ms during the inspiratory phase, after the point of zero flow. The mean inspiratory transpulmonary pressure (PI_tp) was also calculated, by subtraction of the mean inspiratory mouth pressure from during the inspiration.

In the patients with COPD we also estimated the threshold pressure imposed by the dynamic intrinsic positive end-expiratory pressure (PEEPi_dyn). However, this pressure may represent the combined effects of dynamic pulmonary hyperinflation and abdominal muscle activity (21). The latter is reflected by a rise of Pgas during expiration followed by a fall at the beginning of the inspiratory effort (21). Therefore PEEPi_dyn was measured as the amount of negative deflection of preceding the start of inspiratory flow, i.e., the point of the zero flow. Although the accurate method to correct PEEPi_dyn is still debated (21), we corrected PEEPi_dyn for the expiratory muscle activity, if necessary, using a method in line with previous investigators (21) by subtracting from its value the fall of Pgas during the time interval from the onset of inspiratory effort to the point of zero flow.

From the above data, we obtained P_0.1/PI_max and /Pes_max. From these, we computed the tension-time index based on the esophageal pressure, TTI (/Pes_max × TI/Ttot), where PI_max and Pes_max represent the values obtained at rest before the exercise test. From the P_0.1/PI_max ratio we also calculated an index labeled TT_0.1, which is given by P_0.1/ PI_max × TI/Ttot.

Estimation of TTI from TT_0.1

In both groups we determined the individual linear regression equation of TT_0.1 to TTI. This was done for each subject by using the data obtained at rest and at each exercise workload. The mean intercepts and slopes of these linear correlations were calculated for each group. We used these mean values to establish in each group an equation of TTI from TT_0.1. Using this equation we calculated for each subject the estimated TTI (TTI_estim) using the actual TT_0.1 value at rest, at a workload corresponding to approximately 50% of maximum (0.5max) and at maximal exercise (max). The second group of five young normal subjects underwent a similar incremental exercise test. In this group we estimated TTI prospectively using the estimating equations determined on the first group of normal subjects, and compared these estimates to the corresponding experimental values. In four of these subjects a second exercise test was made within a 1-wk period to assess the reproducibility of the results.

Statistics

Values are expressed as mean ± SD. The Students t test for unpaired observations was used for between-group comparisons after confirmation of the normality distribution (Kolmogorov-Smirnov test) and the equality of variance (Levene median test). When these conditions were not met, a Mann-Whitney rank-sum test was used. Linear regression analysis was used to study the relationship between the different noninvasive and invasive pressure parameters (i.e., P_0.1 versus , P_0.1 versus PI_tp, P_0.1 versus  + PEEPi_dyn, P_0.1/PI_max versus / Pes_max, and TT_0.1 versus TTI). Comparison between observed and estimated TTI values was performed by means of the limits of agreement (24). Values of p < 0.05 were considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Respiratory variables at rest, 0.5max, and max obtained in the normal subjects and patients with COPD are shown in Tables 234. In all normal subjects and patients with COPD there was a significant linear correlation (p < 0.02, Table 5) of P_0.1 to and PI_tp, of P_0.1/PI_max to /Pes_max, and of TT_0.1 to TTI. The individual P_0.1 versus  + PEEPi_dyn correlation was also significant in all patients with COPD. Figure 1 shows the relationship between P_0.1 and Pes, and between TT_0.1 and TTI, of a representative normal subject and a patient with COPD. Figure 2 shows the corresponding individual regression lines in all normal subjects and patients with COPD. In the normal subjects the individual regression lines were close to each other. This was also the case for all but one patient with COPD. In this patient the linear correlations of P_0.1 to and of TT_0.1 to TTI were, however, significant (p < 0.01). Therefore, the intersubject slope variability was relatively low in normal subjects compared to patients with COPD. Using all subjects of each group, we calculated the mean intercepts and slopes of the aforementioned linear correlations for the normal and the COPD groups (Table 5). For each group we used these mean values to establish an equation of TTI from TT_0.1. Figure 3 shows in both groups the plot of the difference against the average of estimated and observed TTI values at rest, 0.5max, and max. The limits of agreement are represented by dotted lines. The mean difference of TTI_estim versus TTI was 0.004 ± 0.01 for the normal group and 0.01 ± 0.02 for the COPD group. The limits of agreement of TTI_estim versus TTI ranged from 0.02 to 0.01 for the normal group and from 0.03 to 0.05 for the COPD group. When the aforementioned patient with COPD was removed from the analysis, the patient limits of agreement were tighter (TTI_estim versus TTI, range: 0.03 to 0.03).

View larger version (23K): [in this window] [in a new window]   Figure 1.   Relationship during exercise between P0.1 and Pes in a normal subject (A) and a patient with COPD (B), and between TT0.1 and TTI in the same normal subject (C) and the same COPD patient (D). The thin lines delimit the 95% confidence interval for each regression line.

View larger version (21K): [in this window] [in a new window]   Figure 2.   Individual regression line for the relationship P0.1 versus Pes in all normal subjects (A) and patients with COPD (B), and for the relationship TT0.1 versus TTI in all normal subjects (C) and patients with COPD (D).

View larger version (17K): [in this window] [in a new window]   Figure 3.   Bland and Altman plot (24) between estimated and observed TTI of normal subjects (A) and patients with COPD (B). Dotted line = limits of agreement.

The additional group of five normal subjects also showed significant linear correlations for all of the previously mentioned parameters (i.e., P_0.1 versus , P_0.1 versus PI_tp, P_0.1/ PI_max versus /Pes_max, and TT_0.1 versus TTI; all p < 0.01). In this group, the values of TTI_estim were calculated at rest and at all exercise workloads, using the equations derived from the first group of normal subjects. The limits of agreement ranged from 0.019 to 0.014 for TTI_estim versus TTI. The reproducibility of these results was assessed in four of these normal subjects who performed a second exercise test. Figure 4 shows the individual TT_0.1 versus TTI regression lines obtained during the two tests. Figure 5 shows the limits of agreement for TTI_estim versus TTI obtained during the second test (range: 0.018 to 0.012, Figure 5B), which were similar to those of the first test (Figure 5A), and to those of the first group of normal subjects (Figure 3A).

View larger version (20K): [in this window] [in a new window]   Figure 4.   Individual relationship between TT0.1 and TTI in four normal subjects on two separate exercise tests. Individual regression lines during first test (straight line) and during second test (dotted line). The thin straight or dotted lines correspond to the 95% confidence interval of each regression line.

View larger version (17K): [in this window] [in a new window]   Figure 5.   Bland and Altman plot (24) between estimated and observed TTI of second group of normal subjects (n = 5) during first exercise test (A) and four of these subjects during second exercise test (B). Dotted line = limits of agreement.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The aim of this study was to establish a noninvasive approach to assess the TTI during exercise. In all normal subjects and patients with COPD, P_0.1 was linearly related to during exercise. In each normal subject and each patient with COPD, we showed that a noninvasive tension-time index (TT_0.1) determined from P_0.1, which uses the P_0.1/PI_max ratio as the "tension" component, was also linearly correlated to TTI. On seven normal subjects and nine patients with COPD, we established an equation for estimating TTI from TT_0.1. The limits of agreement between estimated and observed values were wide in the COPD group but tighter in the normal group. In five additional normal subjects, the limits of agreement between observed and prospectively estimated TTI values were also tight. Furthermore, the results from two repeated tests in normal subjects were found to be reproducible.

A noninvasive tension time index of overall inspiratory muscle activity (TTMUS) has been previously validated by Ramonatxo and coworkers (10) in resting healthy subjects and patients with COPD. In this study, PI was obtained from the measurement of P_0.1 according to the equation proposed by Gaultier and coworkers (11). The pressure needed to inspire a breath, P, is equal to a × t^b where a is a constant corresponding to the value of P at 1 s (cm H₂O), t is time, and b a dimensionless constant representing the shape of the pressure profile. If inspiratory pressure increases linearly with time, b = 1 and PI = a TI/2. For a = 10 P_0.1, PI = 5 P_0.1 × TI (10). However, during exercise, in both healthy subjects and patients with COPD the shape of the inspiratory pressure profile is not linear (14, 15). Indeed, while the pressure-time course has been shown to be almost linear at rest, it exhibits an upward convexity during exercise (14, 15). Assuming that the shape of the pressure profile is linear during exercise and using the equation 5 P_0.1 × TI, the error between PI and would be markedly accentuated over the entire inspiratory time. However, in the present study we found a linear relationship between P_0.1 and , and hence between TT_0.1 and TTI. This reflects the fact that, at least in normal subjects, both P_0.1 and increase proportionately during exercise. Based on a complex but comprehensive model analysis that took into account most of the mechanical and neural factors involved in the control of breathing, Younes (25) computed the increase in that would reproduce the average ventilatory output observed in normal subjects at different levels of exercise. According to his computations, /Pes_max should increase curvilinearly with increasing exercise level. Our measurements of noninvasive in the normal subjects closely fit this prediction. In our subjects, the relationship of P_0.1 and P_0.1/PI_max to was also curvilinear with augmenting . Our values of P_0.1 during exercise were superimposed on the relationship previously reported by Hussain and coworkers for normal subjects during exercise (16). Because in normal subjects P_0.1 and increase proportionately with increasing , a linear relationship of P_0.1 to is found. Accordingly, at least in normal subjects, TTI can be assessed noninvasively based on measurements of the mouth occlusion pressure. In this connection it should be noted that Hussain and coworkers (16) have also shown that in healthy subjects during exercise P_0.1 reflects the changes in activity of the major inspiratory muscles.

In each normal subject, we found a significant linear correlation between P_0.1 and during exercise (Figure 1, Table 5). All the parameters derived from P_0.1 were also significantly correlated to the corresponding invasive parameter. A good agreement without systematic bias was found between estimated and measured TTI in the seven normal subjects (Figure 3) when we used the estimating equations established on the same subjects (Table 5). Similar results were obtained when the aforementioned analysis was made prospectively on a limited sample of five additional normal subjects (Figure 5). These results are in line with several studies that have assessed the ability of mouth pressure to reflect esophageal pressure in healthy subjects. Hamnegard and coworkers (26) showed that P_0.1 is closely related to the maximal rate of change in esophageal pressure in normal subjects during CO₂ rebreathing. Yan and coworkers (27) demonstrated the accuracy of twitch mouth pressure (Paw_tw) to predict twitch esophageal pressure (Pes_tw) after electrical phrenic stimulation. Using magnetic bilateral phrenic stimulation (28) or cervical magnetic stimulation (29), it has also been shown that Pes_tw can accurately be predicted from Paw_tw during inspiratory maneuvers, i.e., when the diaphragm is not relaxed. As will be discussed subsequently, in all these studies the main explanation is that normal subjects show a sufficiently short time constant to allow equilibration between alveolar and mouth pressure. This is unlike patients with COPD, who present impaired respiratory mechanics (i.e., airway obstruction, dynamic hyperinflation, slow and imperfect equilibration of mouth with esophageal pressure).

In a small group of normal males we found a reproducible relationship between noninvasive and invasive measurements of TTI (Figure 4), and a close agreement for the estimation of TTI on repeated tests (Figure 5). Although based on a limited sample, our results are encouraging, suggesting that our noninvasive method for measuring TTI can be used to assess inspiratory muscle activity repeatedly over time, at least in young healthy subjects. However, for this method to be generalized to the whole normal population, the accuracy of our estimating equations should be verified with respect to age, sex, and weight. Finally, it has been recently pointed out that there is a need for noninvasive methods to study inspiratory muscle fatigue after high-intensity exercise (30). The usefulness of our method could therefore be assessed in this context.

In patients with COPD, however, we found a wider variability in the relationship between invasive and noninvasive parameters (Figure 2 and Table 5). As a consequence, the estimating equation of TTI from TT_0.1 resulted in weak agreement between estimated and observed data in the COPD group (Figure 3). This variability in the versus P_0.1 slope among patients means that for a given P_0.1 value, may differ among patients with COPD. It is possible from several studies to explain these results. Indeed, several investigators have reported differences between mouth occlusion pressure and iso-time esophageal pressure in patients with COPD (31). This has been attributed to slow and imperfect equilibration of mouth with esophageal pressure (31, 32) as a result of airways obstruction and upper airways compliance (32). Marazzini and coworkers (31) attributed this difference to a delay in equilibration of pressure within the airways, because of lung units with differing time constants. Murciano and coworkers (32) argued that this difference was likely to be due to compliance of the upper airways in severe intubated patients with COPD. Elliott and coworkers (33) also argued that the time delay in the fall of esophageal pressure from end-expiratory esophageal pressure is variable among patients with COPD, and therefore P_0.1 is measured at different times after the activation of inspiratory muscles. The P_0.1 can also be partially produced by the relaxation of expiratory muscles (34). The difference between inspiratory muscle contraction and expiratory muscle relaxation cannot be inferred from mouth pressure changes, and this therefore represents a potential additional factor influencing the versus P_0.1 relationship. In addition to these factors linked to impaired respiratory mechanics, Elliott and coworkers (33), who found P_0.1 to be higher than Pes_0.1 in some patients with severe COPD, proposed a glottis closure while patients exerted a negative pressure with pharyngeal or cheek muscles, which may have explained this difference.

Other researchers have also examined the relationship between mouth and esophageal pressure in patients with COPD. Using electrical or magnetic phrenic stimulation in patients with severe COPD, two studies have been able to assess the ability of Paw_tw to predict Pes_tw. Similowski and coworkers (35) showed that by superimposing bilateral nerve stimulation upon inspiratory muscle efforts, i.e., when the diaphragm is not relaxed, there was a good correlation between Paw_tw and Pes_tw. They concluded that it is possible to record at the mouth an index closely reflecting the effect of diaphragmatic contraction on pleural pressure. Recently, however, Topeli and coworkers (36) reported more conflicting results in a group of patients with severe COPD in whom magnetic stimulation of the phrenic nerves was delivered during gentle inspiratory efforts. They found a strong correlation between Paw_tw and Pes_tw, but wide limits of agreement, and concluded that making predictions of Pes_tw from Paw_tw was unreliable. Nevertheless, by looking at their results (see Figure 4 in Reference 36) it is possible to distinguish in their patients two different subgroups. For one subgroup (11 patients) the mean difference between Paw_tw and Pes_tw appears to be close to zero (cm H₂O), and the limits of agreement seem to be tight. For the second subgroup of two patients, it is possible to deduce a strong correlation between Paw_tw and Pes_tw but a mean difference of approximately 5 cm H₂O. Interestingly, our results are in line with the latter study. Indeed, the estimating equation from noninvasive data was not reliable to predict the index given by the invasive method when we considered the entire COPD group. However, from individual versus P_0.1 or TTI versus TT_0.1 relationships, shown in Figure 2, it can easily be seen that one of the nine patients had a clearly different slope compared with the other patients. For this patient with COPD, we did not find a clear explanation for this difference. He had no other disease such as abdominal obesity, kyphoscoliosis, or pachypleuritis that would have explained his particular pattern of pressure, and the severity of his airway obstruction was not different from the mean group value. In fact this patient showed a versus P_0.1 relationship closer to that of the normal group. When we removed this subject from the calculations, we had narrower limits of agreement and, therefore, a more reliable estimation. In their discussion, Topeli and coworkers (36) suggested that in some patients with COPD, inefficient inspiratory action of the diaphragm, variability in rib cage distortion and variable recruitment of rib cage muscles may explain the different relationships of mouth to esophageal and transdiaphragmatic pressure. Therefore, from the present study and from the study of Topeli and coworkers (36), it appears that in patients with COPD a single equation cannot be used to estimate invasive parameters from mouth pressure.

In conclusion, this study presents a noninvasive TTI during exercise, based on the measurement of P_0.1. In the patients with COPD the agreement was too poor to allow satisfactory estimation of TTI based on TT_0.1. In contrast, in young healthy subjects the changes in TT_0.1 during exercise reflect the changes in TTI. Accordingly, TTI, which requires invasive measurements, may be estimated from mouth pressure measurements in such a population.