Diagnostic accuracy of metronome-paced tachypnea to detect dynamic hyperinflation

Summary

Introduction

This prospective study was carried out to investigate if metronome-paced tachypnea (MPT) can serve as an accurate diagnostic tool to identify patients with chronic obstructive pulmonary disease (COPD) who are susceptible to develop dynamic hyperinflation during exercise. Commonly, this is assessed by measuring change in inspiratory capacity (IC) during cardiopulmonary exercise testing (CPET), which, however, is complex and laborious.

Methods

Fifty-three patients with COPD (FEV₁ 58 ± 22%pred) and 20 age-matched healthy subjects were characterized by lung function testing and performed CPET (reference standard) and MPT. The repeatability coefficient of IC (10·2%) was used as cut-off to classify subjects as hyperinflators during CPET. Subsequently, dynamic hyperinflation was measured after MPT. With receiver operating characteristic analysis, the optimal cut-off for MPT-induced dynamic hyperinflation was determined and sensitivity and specificity of MPT to identify hyperinflators were evaluated.

Results

With 10·2% decrease in IC as cut-off for CPET-induced dynamic hyperinflation, the optimal cut-off for MPT was 11·1% decrease in IC. Using these cut-offs, MPT had a sensitivity of 85% and specificity of 85% to identify the subjects who hyperinflated during CPET.

Conclusions

The MPT test shows good overall accuracy to identify subjects who are susceptible to develop dynamic hyperinflation during CPET. Before considering the use of MPT as a screening tool for dynamic hyperinflation in COPD, sensitivity and specificity need further evaluation.

Introduction

Chronic obstructive pulmonary disease (COPD) is a progressive, treatable disease characterized by not fully reversible airflow limitation. The main feature is airway inflammation, induced by noxious particles like smoke or burning fumes. The airway inflammation leads to structural changes in the airway walls and loss of elasticity of the lung parenchyma (Rabe et al., 2007). These changes have important consequences for the expiratory airflow. During exercise, limitation of expiratory airflow may induce dynamic hyperinflation, an increase in end-expiratory lung volume that is associated with exercise limitation in COPD (O'Donnell, 2006).

Measurements of dynamic hyperinflation are commonly taken during cardiopulmonary exercise testing (CPET). CPET, however, is complex and laborious and only performed in a clinical setting. The development of dynamic hyperinflation, especially in patients with mild and moderate COPD (Ofir et al., 2008), therefore may go unnoticed until considerable exercise limitations occur. With an accurate simple screening tool to detect dynamic hyperinflation in patients with COPD early interventions, such as optimal bronchodilation or exercise training, can be offered to allow patients to continue their activities and prevent deconditioning (Casaburi, 2009). Such a simple surrogate to exercise testing might be metronome-paced tachypnea (MPT). Gelb et al. showed that dynamic hyperinflation induced by breathing for 20 s at twice the resting breathing rate was similar to dynamic hyperinflation after maximal exercise testing in 16 patients with moderate-to-severe COPD (Gelb et al., 2004). In subsequent studies, MPT was used to investigate lung volume responses to bronchodilator use (Gelb et al., 2007, 2009) and the behaviour of dynamic hyperinflation during 2-year follow-up (Hannink et al., 2010a). However, besides its value as study method, this method has not yet been validated for diagnostic use in a larger group of patients. To use MPT as a diagnostic tool, a cut-off value needs to be determined to classify patients as hyperinflators or non-hyperinflators and diagnostic accuracy of the test needs to be established.

The purpose of this study was to investigate the diagnostic accuracy of MPT to detect dynamic hyperinflation. Using CPET as reference standard, subjects were classified as hyperinflator or non-hyperinflator. Because in literature there is no consensus regarding a clinically relevant cut-off value, we used a statistically determined cut-off. Subsequently, our primary aim was to evaluate the sensitivity and specificity of MPT as a screening tool to identify subjects who develop dynamic hyperinflation. In addition, the reliability of measurements during MPT and agreement between MPT- and CPET-induced dynamic hyperinflation were investigated.

Methods

Subjects

For this prospective study, 60 patients with COPD were recruited during their initial screening for pulmonary rehabilitation at the University Centre for Chronic Diseases Dekkerswald in Groesbeek, the Netherlands. Inclusion criteria were age ≥ 40 years, clinically stable condition (no history of exacerbation for the previous 4 weeks) and a diagnosis of COPD, based on a post-bronchodilator forced expiratory volume in one-second/forced vital capacity ratio (FEV₁/FVC) < 70% (Rabe et al., 2007). In addition, 22 age-matched healthy subjects were recruited in patients' and investigators' immediate environment. Exclusion criteria for healthy subjects were FEV₁/FVC < 70%, history of asthma or other lung disease or history of cardiovascular disease. The study was conducted according to the Declaration of Helsinki and was approved by the medical ethical committee of our hospital. All participants gave their informed consent before entering the study.

Study design

All subjects underwent pulmonary function tests and CPET. Based on CPET, subjects were classified as hyperinflators or non-hyperinflators. Subsequently, subjects performed an MPT test to evaluate the accuracy of MPT to identify hyperinflators. Additionally, reliability of the MPT test (repeatability of IC measurements) and agreement between CPET-induced dynamic hyperinflation and MPT-induced dynamic hyperinflation were evaluated.

Pulmonary function tests

Spirometry was performed with a MasterScreen-PFT (Jaeger^®; CareFusion GmbH, Hoechberg, Germany) (Miller et al., 2005). Static lung volumes were assessed by body plethysmography (Jaeger^®; MasterScreen Body, CareFusion GmbH) (Wanger et al., 2005). All measurements were performed according to the American Thoracic Society/European Respiratory Society guidelines for lung function measurements (Miller et al., 2005; Wanger et al., 2005). Reference equations for the calculation of predicted values were those produced by the European Community for Steel and Coal (Quanjer et al., 1993) and predicted values for inspiratory capacity (IC) were calculated as predicted total lung capacity (TLC) minus predicted functional residual capacity (FRC).

Severity of disease was classified according to the Global initiative for Chronic Obstructive Lung Disease (GOLD) stages (Rabe et al., 2007).

Cardiopulmonary exercise test

All subjects performed a symptom-limited incremental exercise test using an electrically braked cycle ergometer (Jaeger^®; Masterlab, Wurzburg, Germany). Subjects wore a leakage-free face mask with a turbine flow transducer and a gas sampling tube (Jaeger^®; Triple V, CareFusion GmbH) attached to it. These were connected to an automated metabolic measuring system (Jaeger^®; Oxycon Pro, CareFusion GmbH).

Measurements were performed according to the American Thoracic Society/European Respiratory Society guidelines for CPET (2003). Reference equations for the calculation of predicted values were those produced by Wasserman (Wasserman et al., 2005).

Given that changes in end-expiratory lung volume cannot be measured directly during exercise, dynamic hyperinflation was estimated by measuring changes in IC (O'Donnell & Webb, 1993). While the subjects sat upright and relaxed on the bike, baseline IC (IC_restCPET) was determined by measuring three maximal inspiratory manoeuvres from a position of passive end-tidal expiration. Instruction to inspire maximally was given after at least four consistent end-expiratory levels, monitored by the real-time display of tidal breathing. The mean of three acceptable manoeuvres was taken as baseline IC (IC_restCPET) (Miller et al., 2005). After three minutes of unloaded pedalling, the load was increased by 5–25 watts per minute, depending on the predicted maximal oxygen consumption (VO₂ max) of the subject, calculated by formulas of Wasserman (Wasserman et al., 2005). At peak exercise, IC was measured (IC_peak) and dynamic hyperinflation (ΔIC_CPET) was calculated as the difference between IC_peak and IC_restCPET.

Metronome-paced tachypnea test

Subjects were seated, breathing through a mouthpiece connected to the spirometer. After a quiet and stable breathing pattern was attained, baseline IC (IC_restMPT) was determined by taking the mean of three acceptable manoeuvres (Miller et al., 2005). Then, a metronome was set at twice the resting breathing rate and subjects were asked to breathe at this pace for 20 s, immediately followed by a maximal inspiratory manoeuvre (IC_MPT) (Gelb et al., 2004). To test reliability of IC_MPT, the procedure was repeated after subjects had returned to their resting breathing level. In line with the criteria for baseline IC manoeuvres, mean IC_MPT was calculated from two acceptable manoeuvres, within 10% of each other. MPT-induced dynamic hyperinflation (∆IC_MPT) was calculated as the difference between mean IC_MPT and IC_restMPT.

Statistical analyses

Characteristics of the participants are described by mean ± standard deviation. Differences between healthy subjects and patients with COPD were tested with unpaired Student's t-tests. To test differences within groups, paired Student's t-tests were used. CPET was used as the reference standard to detect dynamic hyperinflation. Because there is no consensus about the cut-off values for this classification, in this study the repeatability coefficient of the IC measurements at rest was used. When the decrease in IC exceeded the repeatability coefficient, subjects were classified as hyperinflators. A receiver operating characteristic (ROC) analysis was used to determine the optimal cut-off value for MPT-induced dynamic hyperinflation. Subsequently, the sensitivity and specificity of MPT to identify hyperinflators and non-hyperinflators with this cut-off was assessed.

The reliability of IC measurements was assessed with repeated measures ANOVA. The thereby obtained error variance was used to calculate the repeatability coefficient with the following formula: √(2)*1·96*√(mean square error) (Bland & Altman, 1996). Variability in repeated IC manoeuvres was further evaluated by calculation of the coefficient of variation (CoV). To that end, the individual standard deviation of IC measurements was divided by the individual mean.

Agreement between CPET-induced dynamic hyperinflation and MPT-induced dynamic hyperinflation was evaluated using a Bland-Altman plot, illustrating the difference in amount of ∆IC measured by both methods in each subject (Bland & Altman, 1986).

Statistical analyses were performed with spss, version 16.0 (SPSS, Chicago, IL, USA). Statistical significance was set at P<0·05.

Results

Subject characteristics

Figure 1 shows the flow chart illustrating the number of subjects in each stage. In five patients, CPET was not performed because partial pressure of arterial oxygen < 8·0 kPa at rest. Two other patients were excluded from final analyses because CPET results were inconclusive. Characteristics of the study group are displayed in Table 1. Patients (N = 53) and healthy subjects (N = 20) were comparable in age, sex, weight and height. The COPD group consisted of mild to very severe patients (GOLD I, N = 10; GOLD II, N = 22; GOLD III, N = 15; and GOLD IV, N = 6). Fifty-one of the 53 patients were below the lower limit of normal for FEV₁/FVC (Quanjer et al., 1993).

Figure 1. Flow chart illustrating the number of subjects and reasons for exclusion from final analysis.

Table 1. Subject characteristics

	Healthy (N = 20)	COPD (N = 53)
Mean ± SD. BMI, body mass index; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity; TLC, total lung capacity; IC, inspiratory capacity calculated as difference between TLC and intrathoracic gas volume; RV, residual volume; *P<0·05 compared with healthy.
Age, years	57 ± 5	60 ± 8
Sex, M/F	6/14	28/25
Weight, kg	75 ± 14	80 ± 21
Height, cm	171 ± 9	170 ± 9
BMI, kg m^−2	25·7 ± 3·9	27·4 ± 6·3
FEV1, L	3·3 ± 0·7	1·6 ± 0·6*
FEV1, % pred	117 ± 12	58 ± 22*
FEV1/FVC, %	79 ± 5	45 ± 14*
TLC, L	6·5 ± 1·3	7·0 ± 1·5
TLC, % pred	111 ± 10	116 ± 17
IC, % pred	110 ± 23	84 ± 19*
RV/TLC, %	35 ± 4	51 ± 9*

Reliability

The repeated manoeuvres of either IC_rest or IC_MPT were not significantly different. Mean values are shown in Table 2. CoVs (Table 3) were independent of IC_rest (P>0·05), but the variation between repeated manoeuvres correlated with IC_rest (r = 0·23 for IC_restMPT and r = 0·30 for IC_restCPET, P<0·05). In patients with COPD, ΔIC appeared to be higher (more negative) in patients with high IC_rest (r = −0·29 for MPT and r = −0·46 for CPET, P<0·05). With ΔIC expressed as relative change of IC, this correlation disappeared and a more homoscedastic scatter was found (Fig. 2a,b).

Figure 2. (a) Absolute change in IC (∆IC_CPET) in relation to IC_restCPET in patients with COPD shows a heteroscedastic scatter. ∆IC_CPET correlates significantly with IC_restCPET (r = −0·46, P = 0·001). (b) Relative change in IC (∆IC_CPET) in relation to IC_restCPET in patients with COPD shows a homoscedastic scatter. ∆IC_CPET and IC_restCPET are not correlated (r = −0·11, P = 0·446).

Table 2. Inspiratory capacity under different conditions

	Healthy (N = 20)	COPD (N = 53)
Mean ± SEM. ∆IC, change in inspiratory capacity (=dynamic hyperinflation); CPET, cardiopulmonary exercise testing; ICrestMPT, IC at rest before MPT; ICrestCPET, IC at rest before CPET; ICMPT, IC during MPT; ICpeak, IC at peak exercise; MPT, metronome-paced tachypnea. *P<0·05 compared with rest; **P<0·05 compared with healthy; ***P<0·05 compared with MPT.
MPT
ICrestMPT, l	3·12 ± 0·19	2·69 ± 0·10**
ICMPT, l	2·92 ± 0·18*	2·16 ± 0·10*^,**
∆ICMPT, %	−6·7 ± 1·1	−20·3 ± 1·6**
CPET
ICrestCPET, l	3·03 ± 0·17	2·60 ± 0·09**^,***
ICpeak, l	3·08 ± 0·16	2·08 ± 0·08*^,**
∆ICCPET, %	2·6 ± 2·3***	−19·8 ± 1·7**

Table 3. Coefficients of variation for measures of inspiratory capacity

	Healthy (N = 20)	COPD (N = 53)
Mean ± SD. CPET, cardiopulmonary exercise testing; ICrestMPT, inspiratory capacity at rest before MPT; ICrestCPET, IC at rest before CPET; ICMPT, IC during MPT; MPT, metronome-paced tachypnea; *P<0·05 compared with ICrestMPT; **P<0·05 compared with healthy.
MPT
ICrestMPT	2·2 ± 1·6%	3·0 ± 1·5%
ICMPT	2·8 ± 1·8%	4·5 ± 3·1%**
CPET
ICrestCPET	3·1 ± 1·3%*	3·4 ± 2·8%

Repeatability coefficients for IC_restMPT, ΔIC_MPT and IC_restCPET were 8·5%, 12·4% and 10·2%, respectively (corresponding with 0·28 l, 0·30 l and 0·28 l).

Diagnostic accuracy

The cut-off value for dynamic hyperinflation was based on the repeatability coefficient of IC_restCPET, which was 10·2%. Based on that decision threshold (∆IC < −10·2%), 47/73 subjects were classified with CPET as hyperinflators (1 healthy subject and 46 patients; 6/10 GOLD I, 20/22 GOLD II, 14/15 GOLD III, 6/6 GOLD IV).

Receiver operating characteristic curve analysis showed that the optimal cut-off value for MPT-induced dynamic hyperinflation was −11·1% (Fig. 3). Area under the curve (AUC) was 0·912 (95% CI 0·845–0·978, P<0·001). Using 11·1% decrease in IC as cut-off for MPT-induced dynamic hyperinflation in the whole study population, MPT has a sensitivity of 85% and a specificity of 85% to identify subjects who develop dynamic hyperinflation during exercise. In Table 4, the results of the cross-tabulation, comparing classification by CPET and MPT, are described for all subjects and for healthy subjects and patients separately. In the subgroup of patients with COPD, MPT has a sensitivity of 87% and a specificity of 71% to identify the patients who develop dynamic hyperinflation during CPET. Figure 4 depicts the correctly and incorrectly identified hyperinflators and non-hyperinflators for both healthy subjects and different GOLD stages.

Figure 3. Receiver operating characteristic curve plotting the probability of a true positive (y-axis, sensitivity) against the probability of a false positive (x-axis, 1-specificity) for all possible values of the cut-off point (reference cut-off: ΔIC_CPET < −10·2%). Area under the curve (AUC) = 0·912 (95% CI 0·845–0·978). Dotted line: AUC = 0·5. Arrow indicates optimal cut-off value for ΔIC_MPT.

Figure 4. Subjects correctly and incorrectly identified by MPT. Dynamic hyperinflation was measured with CPET (x-axis) and MPT (y-axis). CPET was used as reference standard with 10·2% decrease in IC as decision threshold for classification as hyperinflator. For MPT-induced dynamic hyperinflation, the cut-off was 11·2% decrease in IC. Cut-offs are represented by the dotted lines.

Table 4. Results of the cross-tabulation, comparing the results of classification by CPET and MPT

	All subjects		Healthy		COPD
	CPET		CPET		CPET
	ΔIC < −10·2%	ΔIC > −10·2%	ΔIC < −10·2%	ΔIC > −10·2%	ΔIC < −10·2%	ΔIC > −10·2%
	+	−	+	−	+	−
CPET, cardiopulmonary exercise testing; MPT, metronome-paced tachypnea; ΔIC, change in inspiratory capacity.
MPT
ΔIC < −11·1%  +	40	4	0	2	40	2
ΔIC > −11·1%  −	7	22	1	17	6	5
Sensitivity	85%		0%		87%
Specificity		85%		90%		71%

Agreement

ΔIC_MPT was −16·6 ± 1·4% (−0·44 ± 0·04 l) and ΔIC_CPET was −13·6 ± 1·8% (−0·37 ± 0·05 l). Thus, for relative ΔIC the bias between methods was −3·0%, with 95% CI from −5·8% to −0·1% (absolute bias −70 ml, 95% CI −148 ml; 8 ml). Figure 5 shows the Bland–Altman plot, illustrating the difference in amount of ∆IC measured by both methods in each subject. Limits of agreement were −26·9% to 21·0%.

Figure 5. Bland–Altman plot comparing relative CPET-induced dynamic hyperinflation with MPT-induced dynamic hyperinflation. Decrease in IC is expressed as a positive value to show the overestimation of MPT compared with CPET. The intersection of the y- and x-axis represents the bias between CPET and MPT in dynamic hyperinflation (−3·0 ± 1·4%). The dotted lines represent the upper and lower limits of agreement (21% and −27%, respectively).

Discussion

In this study, the diagnostic accuracy of MPT to detect dynamic hyperinflation was investigated. The cut-off point for CPET-induced dynamic hyperinflation (the reference standard) was based on the statistically derived repeatability coefficient for IC_rest, namely −10·2%. A minimal decrease in IC of 11·1% appeared to be the optimal cut-off for MPT-induced dynamic hyperinflation. Using these decision thresholds, the MPT test has a sensitivity of 85% and a specificity of 85% to identify subjects as hyperinflators or non-hyperinflators.

Overall accuracy of the MPT test was good, as was shown by the AUC (0·912). With the reference standard (CPET), 47 subjects were classified as hyperinflators and 40 of these 47 hyperinflators (85%) were correctly identified with MPT.

The statistically determined cut-off for the reference standard, used in this study, was based on the calculated repeatability coefficient. Because there is some degree of variability in the measurements of IC, one may assume that a small decrease in IC not necessarily reflects dynamic hyperinflation. The repeatability coefficient gives the minimal change in IC that is considered to be a true (significant) change in IC. For IC_restCPET the repeatability coefficient was 10·2%, implying that for individual patient assessment a minimal decrease of 10·2% is needed to classify a patient as hyperinflator. This cut-off agrees with the cut-off value for dynamic hyperinflation used in earlier studies (10% decrease in IC) (O'Donnell et al., 2001; Colucci et al., 2010). However, at this point there is no consensus on which cut-off value, to define dynamic hyperinflation, is clinically relevant. In several studies, hyperinflators were compared with non-hyperinflators. However, depending on which cut-off value was used (ranging from ∆IC < 0 to ∆IC < −4·5%pred), these studies reported different outcomes for exercise capacity, dyspnoea and physical activity level (Garcia-Rio et al., 2009; O'Donnell et al., 2009; Guenette et al., 2011).

Reliability of the test

The low CoV and repeatability coefficient reflect the reproducibility of the measurements and are in agreement with earlier results (Miller et al., 2005). The repeatability coefficient of IC_restMPT implies that for individual patient assessment a minimal decrease of 8·5% is needed to be considered a relevant (true) change in IC. However, ROC analysis showed that lowering the decision threshold for MPT would negatively influence the value of the test. More false-positive test results would occur, so the probability that a positive test result truly reflects dynamic hyperinflation decreases.

The present results showed a heteroscedastic distribution of (absolute) ΔIC in relation to resting IC in patients with COPD. This confirms earlier studies in which the amount of dynamic hyperinflation has been shown not to increase with the severity of COPD, but rather depends on resting IC values (Dykstra et al., 1999; O'Donnell et al., 2001, 2009; Hannink et al., 2010b). Considering this heteroscedastic distribution, defining dynamic hyperinflation as relative change from resting IC values reduces the chance of underestimating or overestimating the random variation for low or high parameter values (Visser et al., 2010). Especially in mild COPD this may be important, as the risk of wrongfully ascribing exertional dyspnoea in these patients to dynamic hyperinflation could be that underlying heart failure, a common unrecognized comorbidity in patients with COPD (Rutten et al., 2005), is overseen.

Agreement

The equal mean dynamic hyperinflation after CPET and MPT is in line with earlier findings (Gelb et al., 2007, 2009). However, Bland–Altman analysis shows that MPT induces slightly more dynamic hyperinflation than CPET (3%). The ROC curve analysis confirms this, because a higher cut-off for the MPT test than for CPET resulted in better overall accuracy. Limits of agreement, however, are wide and larger than the repeatability coefficient. The MPT test and CPET therefore cannot be used interchangeably to assess the amount of dynamic hyperinflation. Consequently, for individual patient assessment, the use of a single method should be recommended when changes over time are to be evaluated.

Limitations

Several limitations in our study design need to be addressed. The available equipment in our pulmonary function laboratory forced us to use different systems for MPT testing and CPET. Subjects breathed through a mouth piece during MPT and through a face mask during CPET, which might have contributed to the differences between both methods in the amount of dynamic hyperinflation measured. However, it is more likely that the lack of agreement is attributable to the different breathing mechanisms that induce dynamic hyperinflation during either CPET or MPT. It is not unlikely that some subjects actively hyperinflate during MPT. Gelb et al. (2004) and Fujimoto et al. (2007) also found decreased IC after MPT in healthy subjects. In patients, active hyperinflation could attenuate the expiratory flow limitation that is the hallmark of COPD; in healthy, it might decrease work of breathing. The imposed higher breathing rate during MPT forces subjects to at least double their expiratory flow and although expiration is mainly a passive process in healthy subjects, it is not unlikely that expiratory muscle activation is required to double the flow. Using the elastic recoil of the lung by breathing at higher lung volumes to increase expiratory flow might be more comfortable then. Our main goal, though, was to assess accuracy of MPT as a diagnostic tool, independent of a difference in underlying breathing mechanisms.

Finally, there might be a selection bias, because a higher prevalence of dynamic hyperinflation may be expected in patients with COPD referred for rehabilitation. To assure whether test results are also reliable in subjects without dynamic hyperinflation, we included healthy subjects in our study sample. In healthy elderly, dynamic hyperinflation may occur when high levels of minute ventilation are reached (Johnson et al., 1999), but its prevalence is expected to be low and in this sample only one healthy subject showed DH.

Clinical relevance

The overall accuracy and reliability of the MPT test illustrate that it could be a reliable and simple alternative to CPET to assess dynamic hyperinflation. Because MPT is less strenuous for the patient than CPET, it seems suitable for screening purposes, enabling early detection of a susceptibility to develop dynamic hyperinflation.

Recent studies reported the occurrence of dynamic hyperinflation in mild COPD (Ofir et al., 2008), as well as a reduction in physical activity in patients with moderate COPD that cannot be explained by clinical characteristics only (Watz et al., 2009). These findings support the concept that it may be important to detect the presence of dynamic hyperinflation during exercise. Further studies are needed, however, to investigate if interventions to reduce dynamic hyperinflation (Casaburi, 2009) can prevent deterioration in physical activity level and patient-related outcomes such as dyspnoea and quality of life in these patients.

At this point, the primary goal was to investigate the accuracy of MPT as a screening tool to detect a susceptibility to dynamic hyperinflation in subjects with COPD. However, one should keep in mind that discrimination based on a fixed cut-off value does not take the actual amount of dynamic hyperinflation into account and only indicates whether dynamic hyperinflation occurs or not.

It might be that other cut-off values prove to be more clinically relevant. Some degree of dynamic hyperinflation may be beneficial (attenuating expiratory flow limitation) as long as a certain inspiratory reserve volume remains. Still, the occurrence of dynamic hyperinflation can be seen as an early sign of deterioration, eventually leading to mechanical constraints.

Conclusion

The current results indicate that MPT shows good overall accuracy to identify subjects who are susceptible to develop dynamic hyperinflation during CPET. The measurements can be performed reliably as was shown by the low CoV and repeatability coefficients. The lack of agreement between measurements with MPT and CPET, though, does not allow both methods to be used interchangeably.

Before considering the use of MPT as a screening tool for dynamic hyperinflation in COPD, sensitivity and specificity need to be evaluated in more general COPD populations. Also, the clinical relevance of cut-offs for dynamic hyperinflation needs further investigation.

Acknowledgements

The authors are indebted to Dr Rogier Donders for his guidance in the statistical analyses. We also thank the staff of the Pulmonary Function Department from the University Centre for Chronic Diseases Dekkerswald for their excellent assistance.

The study was funded by the Netherlands Asthma Foundation (project grant no. 3.4.08.040).

Conflict of interest

All authors declare that they have no conflicts of interest to disclose.