Respiratory Physiology & Neurobiology 207 (2015) 7–13
Contents lists available at ScienceDirect
Respiratory Physiology & Neurobiology journal homepage: www.elsevier.com/locate/resphysiol
Reliability of peak O2 uptake and O2 uptake kinetics in step exercise tests in healthy subjects Paulo de Tarso Müller a,∗ , Gustavo Christofoletti a , Alessandro Moura Zagatto b , Fernanda Viana Paulin a , J Alberto Neder c a Laboratory of Respiratory Pathophysiology (LAFIR), Federal University of Mato Grosso do Sul (UFMS), Rua Filinto Müller S/N, Vila Ipiranga, CEP:79080-090, Campo Grande, Mato Grosso do Sul, Brazil b Faculty of Sciences, Physical Education Department,UNESP—Univ Estadual Paulista, Rua Edmundo Carrijo Coube 14-01–Vargem Limpa, 17.033-360, Bauru, São Paulo, Brazil c Division of Respiratory and Critical Care Medicine, Department of Medicine, Queen’s University and Kingston General Hospital. Richardson House, 102 Stuart Street, Kingston, ON, Canada, K7L 2V6
a r t i c l e
i n f o
Article history: Accepted 1 December 2014 Available online 12 December 2014 Keywords: Aerobic Exercise Kinetics Oxygen consumption Reliability Step tests
a b s t r a c t To date little is known about the reliability of peak oxygen consumption (V˙ O2PEAK ) in incremental metronome paced step tests (IST) and the reliability of on-kinetics V˙ O2 has never been studied. We aimed to study the reliability of both tests. Eleven healthy subjects performed two ISTs until exhaustion. On two different days two duplicate 4 min constant metronome paced step tests (CST) were performed. V˙ O2PEAK , mean response time (MRT) and phase II time constant () were tested for reproducibility using the paired t-tests, in addition to the limits of agreement (LOA) and within subject coefficient of variation (COV). With a 95% LOA of 0.38 to 0.26 L min−1 , −8.7 to 9.1 s and −9.9 to 10.5 s they exhibit a COV of 3%, 4.5% and 6.9% for V˙ O2PEAK , MRT and respectively. ST are sufficiently reliable for maximal and submaximal aerobic power assessments in healthy subjects and new studies of oxygen uptake kinetics in selected patient groups are warranted. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Step tests (ST) elicit vigorous exercise intensities in healthy subjects (Hansen et al., 2011; Petrella et al., 2001; Turner et al., 2004) or selected pulmonary patients (Dal Corso et al., 2013) and are actually increasingly performed in cardiopulmonary research and medical practice (Dal Corso et al., 2007; de Camargo et al., 2011; Fox et al., 2013; Woods et al., 2012). The practical advance of ST is to allow measurement of the valid cardiopulmonary response without requiring expensive ergometer equipment. However, it has been underutilized and there is a major gap in its applicability due to the lack of well-defined protocols and scarce knowledge about ST reproducibility for measuring aerobic power (Petrella et al., 2001), including V˙ O2 kinetics. Such analysis is important in order to determine the effects of training and rehabilitation (Jones and Burnley, 2009), heart transplantation
∗ Corresponding author. Tel.: +55 67 33453149; fax: +55 67 33453049. E-mail addresses: [email protected] (P.d.T. Müller), [email protected] (G. Christofoletti), [email protected] (A.M. Zagatto), [email protected] (F.V. Paulin), [email protected] (J.A. Neder). http://dx.doi.org/10.1016/j.resp.2014.12.001 1569-9048/© 2014 Elsevier B.V. All rights reserved.
(Jendzjowsky et al., 2007) and pharmacological effects (Berton et al., 2010), for example. Studies on the reproducibility of V˙ O2 at peak exercise on a cycle ergometer show a coefficient of variation (COV) of generally below 10% (Akkerman et al., 2010) and the kinetics of V˙ O2 showed good results in terms of reproducibility in healthy subjects (Markovitz et al., 2004; Kilding et al., 2005). However, the reproducibility of V˙ O2 at peak in ST is discordant, although, in general, it has shown good results in earlier studies (Jones et al., 1987; Siconolfi et al., 1985). Thus, knowing the reliability of the ST in the evaluation of V˙ O2PEAK and testing its usefulness in measuring the initial kinetics are key factors in proposing it as an alternative modality in exercise physiology. Because the ST is a test that faithfully reproduces more activities of daily living, such as climbing stairs or a slope (as treadmill), we propose, in this study, to research, for the first time as far as we can ascertain from a search in the literature, the validity of the ST for the on-V˙ O2 kinetics. We hypothesized, that the V˙ O2 at peak exercise and the parameters of on-kinetics in the ST are reproducible and easily applied in healthy subjects with the chance of becoming a valid and applicable method for future clinical studies.
8
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
2. Material and methods 2.1. Participants Fourteen non-trained, healthy individuals, recruited from our University Hospital, volunteered for the study. All participants met the following inclusion criteria: aged between 18 and 60 years at entry, not addicted to alcohol or tobacco and without any known disease or current treatment. Participants with joint or musculoskeletal limitations, obesity (even grade I), hypertension, or any other pathology or taking any medication were not included in the study. The consent form was signed by the subjects after an extensive and detailed explanation of the techniques to be applied. After physical examination, all subjects attended on four occasions, over the period of one week, in order to complete the tests. The study was approved by the ethics committee on human research of our institution and is in accordance with the declaration of Helsinki. 2.2. Study design In this test and retest reproducibility study, we initially conducted a pilot study in which several individuals underwent ST to estimate the rate of step increase and the best exercise time interval for the constant cadence step test (4 or 6 min). On the subjects’ first visit, a physical examination and detailed medical history was performed, in addition to a thorough explanation of all the tests. The subjects were scheduled to perform the exercise tests on four different days, over a period of one week. On the first and third visits identical incremental tests were performed at the same time of day (±30 min). On the second and fourth visits two identical tests were applied each day, with a constant cadence, for reproducibility of the kinetics of V˙ O2 . In all tests (IST or CST) the examiners were blinded to previous step test results. Subjects were instructed not to change their physical activity profile during the week of testing, abstain from coffee, chocolate, tea or other stimulants and consume a light meal 2 h before testing. 2.3. Measurements 2.3.1. Step testing protocol The ST was based on the assumptions of the 6-minute walk test of the American Thoracic Society/American College of Chest Physicians (Laboratories, 2002) and on a previous ST study (Dal Corso et al., 2007), standardized by the examiner with incentives every minute (e.g., “you are doing well, continue.”) on a 20 cm high step without support. On the first and third visits a symptom-limited incremental ST (IST) was conducted driven by a signal emitted by a metronome, starting at 106 beats min−1 with increments of two beats every 20 s. Each beat began on a movement of a leg up or down. The test began with the collection of measurements in the sitting position (i.e., measurement of blood pressure (BP), heart rate (HR) and peripheral oximetry (Dixtal DX 2010, São Paulo, Brazil) at rest. Upon assuming the standing position, the participants proceeded with the following test phases: (i) baseline (2 min), (ii) exercise until exhaustion and (iii) recovery (2 min). BP, HR, oximetry and the rate of perceived exertion (RPE), using the 20-point Borg scale (Borg, 1982), were collected at the end of the exercise. On the first visit (IST) the detection of the cadence (beats min−1 ) at peak exercise (i.e., when the subject could not sustain the exercise for more than 10 s) was planned, sufficient to induce subsequent levels of exercise in the constant cadence step test (CST) which was set at 80–90% of maximum cadence. We assumed that there was a load applied to move up or down which was linearly
related to speed (cadence) during the ST, similar to the concept defined by Porszasz et al. (2003) in a study on the treadmill: S × t = m × g × (Vs × t) × f (t)
(1) (W min−1 ),
where, S is the slope work rate m = mass, g = gravitational constant, V is the rate of change in speed (rate in ms−2 ) and f is the time course of sin (˛), in our case, a constant value, unlike the treadmill exercise, where there is an elevation change. Therefore, the load or the power (Eq. (1)) is directly proportional to speed, and if the speed is increased at a set time (every 20 s), the test is a test for incremental setting. The V˙ O2PEAK criterion was based on observation of maximum exertion (which could not be sustained for more than 10 s) and a respiratory quotient above 1.1. The second and fourth visits were designated for CST, obtained with the cadence calculated as 80–90% of the maximum rate of the first IST. The tests were duplicated with an interval of 30 min of rest on each day of testing. In a quiet environment, with only the examiner, an assistant and the subject, the tests, which were preceded by a rigorous and comprehensive explanation of all procedures, were conducted in the following stages, after measurement of vital signs and resting RPE: (i) baseline (3 min, standing), (ii) constant paced exercise for 4 min and (iii) recovery (sitting for 5 min). The same variables of IST were taken at the beginning and end of the period. We chose 4 min and not 6 min for three reasons: (i) previous study have shown that 3 or 4 time constant (TC, i.e., the time when 63% of steady state V˙ O2 value is achieved) is sufficient to determine the fundamental TC (Bell et al., 2001), (ii) an earlier study had indicated that a plateau was obtained at a speed similar to ours in less than 3 min (Jones et al., 1987) and (iii) in the pilot study we observed, after 3 to 4 min, a higher frequency of fatigue and exercise termination, coupled with the fact that up to 12% of the ST individuals in a previous study did not complete 5 min of exercise because of physical exhaustion (Hansen et al., 2011). It was not our goal to establish a level of exercise above or below lactate threshold, but to establish a high speed (80–90% of maximum cadence), in order to optimize the amplitude of the V˙ O2 , provided care is taken to verify the absence of “slow component” within 3 min (Chiappa et al., 2008; Laveneziana et al., 2009; Berton et al., 2010; Vasilopoulou et al., 2012) and taking into account that the fundamental TC does not change significantly with exercise above or below the lactate threshold (Özyener et al., 2001). 2.3.2. Metabolic data collection The exercise tests were performed in the metabolic system model Vmax 229 (Vyasis, USA), breath by breath, calibrated before each test with precision gases (Gases Gama, São Paulo, Brazil) for two reference points ((i) 26% O2 and nitrogen balance, (ii) 16% O2 and 4% CO2 , balanced with nitrogen). Measurement of inspiratory and expiratory gas flow was performed through a bidirectional, low dead space (39 mL) and low resistance (0.85 were included. To characterize the overall kinetic response to exercise, incorporating a value of baseline V˙ O2 (Eq. (2)) and adopting as the average V˙ O2 the last minute preceding the beginning of the exercise, we calculated the MRT = + ı (Kilding et al., 2005). Unlike the cycle ergometer and treadmill, the beginning of ST exercise shows no significant delay in the transmission of signals to adjust the final constant load (delay time) and so we considered this delay as minimal and insignificant in the ST and this was allowed to vary freely in the equation (Eq. (2)). We used least-square regression with 400 iterations to model the best fit curve in this study. All tests were reviewed for the presence of a slow component by certifying the maintenance of the flat profile of the residual plot within 180 s. 2.3.4. Statistical analysis The number of subjects needed for the study was calculated on PASS11 software (PASS11 .NCSS, LLC, Kaysville, Utah, USA) using tests for paired means based on a maximum difference of means to the time constant () of 4.0 s in diurnal and interdian variability studies (Carter et al., 2002; Markovitz et al., 2004) and a mean standard deviation of 4.0 s (1.9 – 6.2 s in the literature; (Özyener et al., 2001; Carter et al., 2002). Thus, we calculated that 12 individuals would ensure a power of 0.812 for ˛ = 0.01. Statistical analysis and modeling to mono-exponential curves, the latter using least-squares non-linear regression analysis to automatically determine the best model, were performed using GraphPad Prism Software (GraphPad Software 5.0, San Diego, CA). All variables except RPE were normally distributed according to the Shapiro–Wilks normality tests. Descriptive analysis [mean ± SD] was used for the anthropometric variables and for calculating the average of the two measures combined on each day of CST. Paired t-tests were used to compare, with a level of significance of 5%, the parameters associated with the IST (test1 and test2 ) and CST (test1 and test2 ) on different days. We used the Pearson coefficient of correlation for association between peak and kinetics parameters of V˙ O2 . The test-retest reproducibility was assessed by limits of agreement (LOA) of 95%, which means calculating the upper and lower limits in which 95% of the differences should occur in the same unit in which they were measured (in our case milliliters or seconds) (Bland and Altman, 1986), and are calculated as the product of 1.96 by the standard deviation of the difference between test1 and test2 . To measure the error we used the following formula: 100 × (LOA/grand mean), where the grand mean represents: mean test1 + mean test2 /2 (Atkinson and Nevill,
9
Table 1 V˙ O2PEAK associated parameters during incremental exercise step test (test1 × test2 ). Values are mean ± SD. (b) = baseline; RPE = rate of perceived exertion; HR = heart rate. Parameter V˙ O2(b) (mL min−1 ) V˙ O2PEAK (mL min−1 ) Total step (n◦ .) RPE HR (bpm)
Test1
Test2
355 ± 75,0 2317 185 14 162
± ± ± ±
603,5 80,0 1,8 17,7
p
355 ± 71,9
0.997
± ± ± ±
0.241 0.465 0.705 0.062
2378 194 14 169
585,6 68,4 2,9 17,8
1998). The coefficient of variation (COV) between individuals was calculated as 100 × (eSD − 1), and SD as the standard deviation of the difference between test1 and test2 , after logarithmic transformation of values, a more precise and robust calculus supported by several authors (Bland and Altman, 1996; Atkinson and Nevill, 1998; Hopkins, 2000). 3. Results 3.1. Subject characteristics All subjects had peripheral arterial saturation higher than 95% at rest, and none had significant exercise desaturation, i.e., SpO2 > 4%. Eleven participants (four women), with a mean ± SD age of 31.2 ± 9.2 years, height of 175.0 ± 4.2 cm, body mass of 88.1 ± 10.6 kg and a body mass index (BMI) of 28.8(3.5) kg m−2 were enrolled in the study. The range of age (21–49 years) was tighter than planned and the range of height (170–182 cm) was casually advantageous for this not height-adjusted step study (Further details, see Section 4.3). Three participants were excluded from the analysis. Two participants were excluded due to evident slow component visualization. One patient was excluded from the analysis by presenting a model with regression coefficients 0.05). All participants were active non-trained healthy individuals, as confirmed by physical examination and individual medical investigations. 3.2. V˙ O2PEAK in IST All major parameters obtained in test1 and test2 (ITD) are presented in Table 1. The HR at end of exercise was corresponding to 85.7% and 89.4% of maximal predicted HR (i.e., estimated using the equation 220-age), respectively. The total number of steps and rate of perceived exertion were not statistically different between test1 and test2 (p > 0.05) and the exercise time ranged from 3 to 16 min, with a mean duration of 9.5 min. 3.3. V˙ O2 kinetics in CST Table 2 shows the comparison of mean ± SD of all parameters obtained in the analysis of the kinetics of V˙ O2 in test1 and Table 2 On-transient kinetics and associated parameters during constant step test exercise (test1 × test2 ). Values are mean ± SD. A = amplitude; ı = time delay; = time constant; MRT = mean response time; HR = heart rate. Parameter V˙ O2(b) (mL min−1 ) V˙ O2 (plateau) (mL min−1 ) A (mL min−1 ) ı (s) MRT(0–3 min) (s) (20 s–3 min) (s) HR(4 min) (bpm)
Test1
Test2
375 ± 68.4 1995 1618 11.6 48.6 36.9 143
± ± ± ± ± ±
330.5 354.5 4.1 8.5 9.7 22.2
p
335 ± 67.3
0.065
± ± ± ± ± ±
0.588 0.730 0.932 0.887 0.865 0.912
1971 1636 11.7 48.4 36.7 143
319.6 329.3 4.4 6.8 7.1 17.0
10
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
Fig. 1. Typical on-transition fit model (A) and two on-transition superimposed to exemplify reproducibility of CST (B), where grey circles represent test1 ( = 45 s; MRT = 51 s) and black circles test2 ( = 41 s; MRT = 48 s) for the same subject. For clarity, exponential model and residuals were not illustrated.
test2 (CST). For all but one participant (excluded) the coefficient of regression of fitted curves was higher than 0.85, with a mean coefficient of r = 0.95 (0.89 − 0.99). Fig. 1 shows an example of a regression model (A) and an example of a rest-to-exercise transition in ST with an overlap of test1 and test2 to CST (B). The analysis of the residuals showed excellent quality models, with residues ranging randomly around the zero error, with the exception of the two excluded subjects. The amplitude was sufficiently high for adequate analysis of the response in the exercise transition. Time delay (ı) showed the greatest measurement error (46%), but in this model, where it varied freely to optimize the goodness of fit of the monoexponential regression, it has no physiological significance (it does not have the same significance as in Phase I).
4.1. V˙ O2PEAK in IST The available studies on reproducibility of V˙ O2PEAK (ST) in healthy people are widely variable as to protocol type. Thus, there are CSTs for 10 min with arm support (Jones et al., 1987), tests of two
3.4. Association between V˙ O2PEAK and on-kinetics parameters We could not find association, in a matrix of correlations, between kinetics parameters and V˙ O2PEAK (test1 and test2 ). 3.5. Reproducibility After confirmation of normal distribution for all parameters using the Shapiro–Wilk test (p > 0.05), we found that the main parameters associated with IST and CST (Tables 1 and 2 respectively) showed no significant difference between tests 1 and 2 (p > 0.05 for all paired comparisons). A qualitative analysis using the Bland–Altman plot for the three main parameters of interest (Fig. 2) was conducted to visually check the dispersion difference between tests 1 and 2 and the grand mean of these tests. The absolute difference between tests 1 and 2 and the grand mean for each transition for V˙ O2PEAK , MRT and showed no statistically significant correlation (p > 0.05 for all results), indicating that there is homoscedasticity distribution for all three parameters (Atkinson and Nevill, 1998). The coefficients of variation, LOA and other variables of reproducibility analysis are shown for all parameters in Tables 3 and 4, where we observe that the V˙ O2PEAK showed excellent reproducibility and MRT was more reproducible than . 4. Discussion This study investigated the reproducibility of an incremental step test and a 4-min constant cadence step test in providing V˙ O2PEAK and parameters of V˙ O2 on-kinetics in healthy subjects. Our results indicate that these tests provide reproducible values for these variables which were well commensurate to those described in the literature for other exercise test modalities.
Fig. 2. Bland–Altman plots for the differences (errors) against the mean for (A) V˙ O2PEAK , (B) MRT and (C) tau (). The bias line (solid) and 95% LOA (dashed) are also shown.
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
11
Table 3 Means and measures of reliability for incremental step tests to peak exercise. LOA = limits of agreements; (b) = baseline. CV = coefficient of variation. Parameters
Mean of test 1 and 2
Mean ± SD difference
95% LOA
Measurement error (%)
Systematic bias (%)
Within subject CV (%)
V˙ O2(b) (mL min−1 ) V˙ O2PEAK (mL min−1)
355
−0.1 ± 81
−159 to 160
45.0
−0.03
9.4
−383 to 260 −90 to 72 −29 to 15
13.7 42.6 13.3
2.6 5.2 4.3
3.0 7.8 3.0
Total step (n◦ .) HRpeak (bpm)
2347 190 165
62 ± 164 10 ± 5 7.1 ± 11
steps (20 cm each) with a free rate (Petrella et al., 2001), the Chester step test (Buckley et al., 2004) and a CST adjusted for age and HR, and the step height dependent on the length of the leg (Thomas et al., 1993). Because of this, comparability in healthy subjects is limited. In addition, different statistical calculations of reproducibility are available. We could not find any reproducibility studies of IST in the conditions of this study, i.e.,without support and with a step height of 20 cm in healthy people. However, a COV (V˙ O2PEAK ) of 10.4% from a modified Canadian Aerobic Fitness Test was reported (Thomas et al., 1993) in a wide age range in healthy subjects, and, coefficients of test–retest correlation of 0.97 for men and 0.98 for women were mentioned in a free cadence ST (Petrella et al., 2001). However, correlation coefficients depend on a higher or lower range of the variable studied (and sample) and detect an “association” between the test–retest variable and not necessarily reproducibility. The mean difference of 61.6 mL (2.6%) between test1 and test2 for V˙ O2PEAK is a physiologically acceptable difference for studies of aerobic power, and coupled with an accuracy of 321 mL for 95% of the values achieved in all tests and a COV of 3%, makes the determination of V˙ O2PEAK in the IST reproducible and applicable in routine aerobic evaluation. A wide study on a normal cycle ergometer found a mean absolute difference between the tests of 42 mL min−1 and a COV of 5.1% (Skinner et al., 1999), a result very similar to this study. 4.2. V˙ O2 on-kinetics in CST The kinetics of the rest-to-exercise transition for V˙ O2 is possibly influenced by the type of ergometer/instrument used for exercise, because different muscle groups and types of muscle fibers may be activated differently (Carter et al., 2000; Koga et al., 2005). The ST has peculiarities that cannot be ignored, such as intense activation of the leg eccentric muscle contraction groups, which are not usually found in the cycle ergometer exercise. The mean fundamental or Phase II of 36.8 s found in our study was similar to the 34.0 s (Özyener et al., 2001), but slower than the 21.6 s (Carter et al., 2000) or 23.9 s (Burnley et al., 2000) found in healthy untrained subjects in cycle ergometry, with a comparable Phase II amplitude (∼1500 mL) and individuals’ characteristics. The differences may be due to the different degrees of physical fitness of the individuals and different mathematical models used, in addition to reasons related to the type of exercise and ergometer. Comparative studies are needed to determine if there are actually differences in V˙ O2 kinetics related to the type of ergometer/instrument. A comparative study using O2 -deficit calculations
suggested marked differences in V˙ O2 on-kinetics between the cycle ergometer and ST, suggesting that these differences would be secondary to (i) type of muscle fibers recruited; (ii) static and dynamic components of muscle contraction and/or (iii) the muscle perfusion (di Prampero et al., 1989). The average MRT of 48.5 s in our study, however, is slower than the average values found in cycle ergometry in populations similar to ours: 43.9 s (Carter et al., 2000) and 44.1 s (Burnley et al., 2000), considering MRT = + ı at a high intensity level of exercise. The main difference is due to a slower Phase II and a shorter ı in our study. The slower average in our study, and in consequence the MRT, could be due to the transition which occurs from rest-to-exercise in ST, unlike most cycle ergometer studies which use a transition from unload-to-exercise. The difference is that in rest-to-exercise there is a greater and possibly more prolonged mobilization pool of blood through the venous return from the legs to the lungs during the cardiopulmonary phase (Phase I) of exercise, which would also explain the longer MRT in the knee extension exercise in comparison to the cycle ergometer (Koga et al., 2005), therefore including some Phase I data in the fitting model (when excluding the first 20 s of data) and, in consequence, reducing ı in such free modeling for ı, unlike ı constrained to Phase I–Phase II transitions (ı = Phase I) as in Carter et al. (2000) and Burnley et al. (2000). The fact that there are no previous references to this modality of exercise was a challenge and limited our ability to compare the reproducibility of V˙ O2 kinetics in ST. Our COV of 6.9% for Phase II and 4.5% for the MRT indicates an acceptable range for application as an exercise test and is within the range of physiological measures in humans and compares, although using different methodologies, to the 11.5% (Markovitz et al., 2004) and 11.0% (Özyener et al., 2001) for TC in untrained subjects on cycle ergometry. Similar to V˙ O2PEAK , there was no significant difference between the means of the two tests (test1 and test2 ), indicating that there was no significant bias due to the effect of learning on the second test, interfering in the results. The accuracy of 10 s for the of Phase II, suggests that for 95% of the time that the subject is tested, it is possible for an individual with, for example, a = 41 s (Fig. 1B) to obtain a new of between 31 s and 51 s. However, as suggested in Fig. 2C, 10 of the 11 subjects tested had an accuracy of 7.0 s. For the MRT, the accuracy was 9s (Table 4), therefore higher than for the Phase II and with a lower COV (4.5%). This finding was also observed in a study of reproducibility in trained individuals, where the MRT was generally more reproducible than Phase II (Kilding et al., 2005) on the treadmill.
Table 4 Means and measures of reliability (on-kinetics) for constant step tests. A = amplitude; ı = time delay; = time constant; MRT = mean response time; HR = heart rate. Parameters V˙ O2(b) (mL min−1 ) V˙ O2 ,(plateau) (mL min−1 ) −1
A (mL min ) MRT(0-3 min) (s) ı (s) (20 s-3 min) (s) HR(4 min) (bpm)
Mean of Test 1 and 2 355 1983 1627 48.5 11.7 36.8 143
Mean ± SD difference
95% LOA
Measurement error (%)
Systematic bias (%)
−39 ± 63
−84 to 164
35.0
−11.1
8.3
−23 17 −0.2 0.1 −0.3 0.4
−249 to 296 −336 to 301 −8.7 to 9.1 −5.5 to 5.4 −9.9 to 10.5 −21 to 21
13.7 19.6 18.3 46.6 27.7 14.5
−1.2 1.1 −0.4 0.9 −0.8 0.3
3.0 4.3 4.5 11.5 6.9 3.4
± ± ± ± ± ±
132.7 154.8 4.3 2.6 5.0 10
Within subject CV (%)
12
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
4.3. Methodological considerations An important point to be considered is the chosen protocol for the transition rest-to-exercise and kinetics analysis, i.e., 80–90% of the maximal speed (step min−1 ) reached in IST, to optimize the signal associated with the beginning of exercise. The objective was to achieve a range of V˙ O2 high enough for a more accurate regression analysis, since a larger amplitude is directly associated with a better regression coefficient (Bell et al., 2001). However, transitions to a sub anaerobic threshold speed (moderate exercise) may be more comfortable for patients with a more limited cardiopulmonary reserve, bearing in mind that there may be a greater noise associated with a lower amplitude, plus a lower V˙ O2 signal-to-noise relation when the rest (as in ST) instead of a “baseline” load is used, as has already been noted in the COV of up to 37% in resting V˙ O2 (Markovitz et al., 2004). This limitation could be partially overcome by a low speed adopted as a “baseline” for the ST. In the incremental test, the adoption of a step of fixed height also imposes certain limitations, since the ideal would be to adjust the height of the step to the subjects’ body height (Hansen et al., 2011). Without this, the result could be very large differences in exercise time, because the load theoretically applied on ST depends on the height of the step (see Eq. (2)). Calculating the maximum speed (step min−1 ) in a difficult situation like this, could result in an exercise that is too long, with a risk of muscle injury or, alternatively, a very short exercise time (e.g.,
Contents lists available at ScienceDirect
Respiratory Physiology & Neurobiology journal homepage: www.elsevier.com/locate/resphysiol
Reliability of peak O2 uptake and O2 uptake kinetics in step exercise tests in healthy subjects Paulo de Tarso Müller a,∗ , Gustavo Christofoletti a , Alessandro Moura Zagatto b , Fernanda Viana Paulin a , J Alberto Neder c a Laboratory of Respiratory Pathophysiology (LAFIR), Federal University of Mato Grosso do Sul (UFMS), Rua Filinto Müller S/N, Vila Ipiranga, CEP:79080-090, Campo Grande, Mato Grosso do Sul, Brazil b Faculty of Sciences, Physical Education Department,UNESP—Univ Estadual Paulista, Rua Edmundo Carrijo Coube 14-01–Vargem Limpa, 17.033-360, Bauru, São Paulo, Brazil c Division of Respiratory and Critical Care Medicine, Department of Medicine, Queen’s University and Kingston General Hospital. Richardson House, 102 Stuart Street, Kingston, ON, Canada, K7L 2V6
a r t i c l e
i n f o
Article history: Accepted 1 December 2014 Available online 12 December 2014 Keywords: Aerobic Exercise Kinetics Oxygen consumption Reliability Step tests
a b s t r a c t To date little is known about the reliability of peak oxygen consumption (V˙ O2PEAK ) in incremental metronome paced step tests (IST) and the reliability of on-kinetics V˙ O2 has never been studied. We aimed to study the reliability of both tests. Eleven healthy subjects performed two ISTs until exhaustion. On two different days two duplicate 4 min constant metronome paced step tests (CST) were performed. V˙ O2PEAK , mean response time (MRT) and phase II time constant () were tested for reproducibility using the paired t-tests, in addition to the limits of agreement (LOA) and within subject coefficient of variation (COV). With a 95% LOA of 0.38 to 0.26 L min−1 , −8.7 to 9.1 s and −9.9 to 10.5 s they exhibit a COV of 3%, 4.5% and 6.9% for V˙ O2PEAK , MRT and respectively. ST are sufficiently reliable for maximal and submaximal aerobic power assessments in healthy subjects and new studies of oxygen uptake kinetics in selected patient groups are warranted. © 2014 Elsevier B.V. All rights reserved.
1. Introduction Step tests (ST) elicit vigorous exercise intensities in healthy subjects (Hansen et al., 2011; Petrella et al., 2001; Turner et al., 2004) or selected pulmonary patients (Dal Corso et al., 2013) and are actually increasingly performed in cardiopulmonary research and medical practice (Dal Corso et al., 2007; de Camargo et al., 2011; Fox et al., 2013; Woods et al., 2012). The practical advance of ST is to allow measurement of the valid cardiopulmonary response without requiring expensive ergometer equipment. However, it has been underutilized and there is a major gap in its applicability due to the lack of well-defined protocols and scarce knowledge about ST reproducibility for measuring aerobic power (Petrella et al., 2001), including V˙ O2 kinetics. Such analysis is important in order to determine the effects of training and rehabilitation (Jones and Burnley, 2009), heart transplantation
∗ Corresponding author. Tel.: +55 67 33453149; fax: +55 67 33453049. E-mail addresses: [email protected] (P.d.T. Müller), [email protected] (G. Christofoletti), [email protected] (A.M. Zagatto), [email protected] (F.V. Paulin), [email protected] (J.A. Neder). http://dx.doi.org/10.1016/j.resp.2014.12.001 1569-9048/© 2014 Elsevier B.V. All rights reserved.
(Jendzjowsky et al., 2007) and pharmacological effects (Berton et al., 2010), for example. Studies on the reproducibility of V˙ O2 at peak exercise on a cycle ergometer show a coefficient of variation (COV) of generally below 10% (Akkerman et al., 2010) and the kinetics of V˙ O2 showed good results in terms of reproducibility in healthy subjects (Markovitz et al., 2004; Kilding et al., 2005). However, the reproducibility of V˙ O2 at peak in ST is discordant, although, in general, it has shown good results in earlier studies (Jones et al., 1987; Siconolfi et al., 1985). Thus, knowing the reliability of the ST in the evaluation of V˙ O2PEAK and testing its usefulness in measuring the initial kinetics are key factors in proposing it as an alternative modality in exercise physiology. Because the ST is a test that faithfully reproduces more activities of daily living, such as climbing stairs or a slope (as treadmill), we propose, in this study, to research, for the first time as far as we can ascertain from a search in the literature, the validity of the ST for the on-V˙ O2 kinetics. We hypothesized, that the V˙ O2 at peak exercise and the parameters of on-kinetics in the ST are reproducible and easily applied in healthy subjects with the chance of becoming a valid and applicable method for future clinical studies.
8
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
2. Material and methods 2.1. Participants Fourteen non-trained, healthy individuals, recruited from our University Hospital, volunteered for the study. All participants met the following inclusion criteria: aged between 18 and 60 years at entry, not addicted to alcohol or tobacco and without any known disease or current treatment. Participants with joint or musculoskeletal limitations, obesity (even grade I), hypertension, or any other pathology or taking any medication were not included in the study. The consent form was signed by the subjects after an extensive and detailed explanation of the techniques to be applied. After physical examination, all subjects attended on four occasions, over the period of one week, in order to complete the tests. The study was approved by the ethics committee on human research of our institution and is in accordance with the declaration of Helsinki. 2.2. Study design In this test and retest reproducibility study, we initially conducted a pilot study in which several individuals underwent ST to estimate the rate of step increase and the best exercise time interval for the constant cadence step test (4 or 6 min). On the subjects’ first visit, a physical examination and detailed medical history was performed, in addition to a thorough explanation of all the tests. The subjects were scheduled to perform the exercise tests on four different days, over a period of one week. On the first and third visits identical incremental tests were performed at the same time of day (±30 min). On the second and fourth visits two identical tests were applied each day, with a constant cadence, for reproducibility of the kinetics of V˙ O2 . In all tests (IST or CST) the examiners were blinded to previous step test results. Subjects were instructed not to change their physical activity profile during the week of testing, abstain from coffee, chocolate, tea or other stimulants and consume a light meal 2 h before testing. 2.3. Measurements 2.3.1. Step testing protocol The ST was based on the assumptions of the 6-minute walk test of the American Thoracic Society/American College of Chest Physicians (Laboratories, 2002) and on a previous ST study (Dal Corso et al., 2007), standardized by the examiner with incentives every minute (e.g., “you are doing well, continue.”) on a 20 cm high step without support. On the first and third visits a symptom-limited incremental ST (IST) was conducted driven by a signal emitted by a metronome, starting at 106 beats min−1 with increments of two beats every 20 s. Each beat began on a movement of a leg up or down. The test began with the collection of measurements in the sitting position (i.e., measurement of blood pressure (BP), heart rate (HR) and peripheral oximetry (Dixtal DX 2010, São Paulo, Brazil) at rest. Upon assuming the standing position, the participants proceeded with the following test phases: (i) baseline (2 min), (ii) exercise until exhaustion and (iii) recovery (2 min). BP, HR, oximetry and the rate of perceived exertion (RPE), using the 20-point Borg scale (Borg, 1982), were collected at the end of the exercise. On the first visit (IST) the detection of the cadence (beats min−1 ) at peak exercise (i.e., when the subject could not sustain the exercise for more than 10 s) was planned, sufficient to induce subsequent levels of exercise in the constant cadence step test (CST) which was set at 80–90% of maximum cadence. We assumed that there was a load applied to move up or down which was linearly
related to speed (cadence) during the ST, similar to the concept defined by Porszasz et al. (2003) in a study on the treadmill: S × t = m × g × (Vs × t) × f (t)
(1) (W min−1 ),
where, S is the slope work rate m = mass, g = gravitational constant, V is the rate of change in speed (rate in ms−2 ) and f is the time course of sin (˛), in our case, a constant value, unlike the treadmill exercise, where there is an elevation change. Therefore, the load or the power (Eq. (1)) is directly proportional to speed, and if the speed is increased at a set time (every 20 s), the test is a test for incremental setting. The V˙ O2PEAK criterion was based on observation of maximum exertion (which could not be sustained for more than 10 s) and a respiratory quotient above 1.1. The second and fourth visits were designated for CST, obtained with the cadence calculated as 80–90% of the maximum rate of the first IST. The tests were duplicated with an interval of 30 min of rest on each day of testing. In a quiet environment, with only the examiner, an assistant and the subject, the tests, which were preceded by a rigorous and comprehensive explanation of all procedures, were conducted in the following stages, after measurement of vital signs and resting RPE: (i) baseline (3 min, standing), (ii) constant paced exercise for 4 min and (iii) recovery (sitting for 5 min). The same variables of IST were taken at the beginning and end of the period. We chose 4 min and not 6 min for three reasons: (i) previous study have shown that 3 or 4 time constant (TC, i.e., the time when 63% of steady state V˙ O2 value is achieved) is sufficient to determine the fundamental TC (Bell et al., 2001), (ii) an earlier study had indicated that a plateau was obtained at a speed similar to ours in less than 3 min (Jones et al., 1987) and (iii) in the pilot study we observed, after 3 to 4 min, a higher frequency of fatigue and exercise termination, coupled with the fact that up to 12% of the ST individuals in a previous study did not complete 5 min of exercise because of physical exhaustion (Hansen et al., 2011). It was not our goal to establish a level of exercise above or below lactate threshold, but to establish a high speed (80–90% of maximum cadence), in order to optimize the amplitude of the V˙ O2 , provided care is taken to verify the absence of “slow component” within 3 min (Chiappa et al., 2008; Laveneziana et al., 2009; Berton et al., 2010; Vasilopoulou et al., 2012) and taking into account that the fundamental TC does not change significantly with exercise above or below the lactate threshold (Özyener et al., 2001). 2.3.2. Metabolic data collection The exercise tests were performed in the metabolic system model Vmax 229 (Vyasis, USA), breath by breath, calibrated before each test with precision gases (Gases Gama, São Paulo, Brazil) for two reference points ((i) 26% O2 and nitrogen balance, (ii) 16% O2 and 4% CO2 , balanced with nitrogen). Measurement of inspiratory and expiratory gas flow was performed through a bidirectional, low dead space (39 mL) and low resistance (0.85 were included. To characterize the overall kinetic response to exercise, incorporating a value of baseline V˙ O2 (Eq. (2)) and adopting as the average V˙ O2 the last minute preceding the beginning of the exercise, we calculated the MRT = + ı (Kilding et al., 2005). Unlike the cycle ergometer and treadmill, the beginning of ST exercise shows no significant delay in the transmission of signals to adjust the final constant load (delay time) and so we considered this delay as minimal and insignificant in the ST and this was allowed to vary freely in the equation (Eq. (2)). We used least-square regression with 400 iterations to model the best fit curve in this study. All tests were reviewed for the presence of a slow component by certifying the maintenance of the flat profile of the residual plot within 180 s. 2.3.4. Statistical analysis The number of subjects needed for the study was calculated on PASS11 software (PASS11 .NCSS, LLC, Kaysville, Utah, USA) using tests for paired means based on a maximum difference of means to the time constant () of 4.0 s in diurnal and interdian variability studies (Carter et al., 2002; Markovitz et al., 2004) and a mean standard deviation of 4.0 s (1.9 – 6.2 s in the literature; (Özyener et al., 2001; Carter et al., 2002). Thus, we calculated that 12 individuals would ensure a power of 0.812 for ˛ = 0.01. Statistical analysis and modeling to mono-exponential curves, the latter using least-squares non-linear regression analysis to automatically determine the best model, were performed using GraphPad Prism Software (GraphPad Software 5.0, San Diego, CA). All variables except RPE were normally distributed according to the Shapiro–Wilks normality tests. Descriptive analysis [mean ± SD] was used for the anthropometric variables and for calculating the average of the two measures combined on each day of CST. Paired t-tests were used to compare, with a level of significance of 5%, the parameters associated with the IST (test1 and test2 ) and CST (test1 and test2 ) on different days. We used the Pearson coefficient of correlation for association between peak and kinetics parameters of V˙ O2 . The test-retest reproducibility was assessed by limits of agreement (LOA) of 95%, which means calculating the upper and lower limits in which 95% of the differences should occur in the same unit in which they were measured (in our case milliliters or seconds) (Bland and Altman, 1986), and are calculated as the product of 1.96 by the standard deviation of the difference between test1 and test2 . To measure the error we used the following formula: 100 × (LOA/grand mean), where the grand mean represents: mean test1 + mean test2 /2 (Atkinson and Nevill,
9
Table 1 V˙ O2PEAK associated parameters during incremental exercise step test (test1 × test2 ). Values are mean ± SD. (b) = baseline; RPE = rate of perceived exertion; HR = heart rate. Parameter V˙ O2(b) (mL min−1 ) V˙ O2PEAK (mL min−1 ) Total step (n◦ .) RPE HR (bpm)
Test1
Test2
355 ± 75,0 2317 185 14 162
± ± ± ±
603,5 80,0 1,8 17,7
p
355 ± 71,9
0.997
± ± ± ±
0.241 0.465 0.705 0.062
2378 194 14 169
585,6 68,4 2,9 17,8
1998). The coefficient of variation (COV) between individuals was calculated as 100 × (eSD − 1), and SD as the standard deviation of the difference between test1 and test2 , after logarithmic transformation of values, a more precise and robust calculus supported by several authors (Bland and Altman, 1996; Atkinson and Nevill, 1998; Hopkins, 2000). 3. Results 3.1. Subject characteristics All subjects had peripheral arterial saturation higher than 95% at rest, and none had significant exercise desaturation, i.e., SpO2 > 4%. Eleven participants (four women), with a mean ± SD age of 31.2 ± 9.2 years, height of 175.0 ± 4.2 cm, body mass of 88.1 ± 10.6 kg and a body mass index (BMI) of 28.8(3.5) kg m−2 were enrolled in the study. The range of age (21–49 years) was tighter than planned and the range of height (170–182 cm) was casually advantageous for this not height-adjusted step study (Further details, see Section 4.3). Three participants were excluded from the analysis. Two participants were excluded due to evident slow component visualization. One patient was excluded from the analysis by presenting a model with regression coefficients 0.05). All participants were active non-trained healthy individuals, as confirmed by physical examination and individual medical investigations. 3.2. V˙ O2PEAK in IST All major parameters obtained in test1 and test2 (ITD) are presented in Table 1. The HR at end of exercise was corresponding to 85.7% and 89.4% of maximal predicted HR (i.e., estimated using the equation 220-age), respectively. The total number of steps and rate of perceived exertion were not statistically different between test1 and test2 (p > 0.05) and the exercise time ranged from 3 to 16 min, with a mean duration of 9.5 min. 3.3. V˙ O2 kinetics in CST Table 2 shows the comparison of mean ± SD of all parameters obtained in the analysis of the kinetics of V˙ O2 in test1 and Table 2 On-transient kinetics and associated parameters during constant step test exercise (test1 × test2 ). Values are mean ± SD. A = amplitude; ı = time delay; = time constant; MRT = mean response time; HR = heart rate. Parameter V˙ O2(b) (mL min−1 ) V˙ O2 (plateau) (mL min−1 ) A (mL min−1 ) ı (s) MRT(0–3 min) (s) (20 s–3 min) (s) HR(4 min) (bpm)
Test1
Test2
375 ± 68.4 1995 1618 11.6 48.6 36.9 143
± ± ± ± ± ±
330.5 354.5 4.1 8.5 9.7 22.2
p
335 ± 67.3
0.065
± ± ± ± ± ±
0.588 0.730 0.932 0.887 0.865 0.912
1971 1636 11.7 48.4 36.7 143
319.6 329.3 4.4 6.8 7.1 17.0
10
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
Fig. 1. Typical on-transition fit model (A) and two on-transition superimposed to exemplify reproducibility of CST (B), where grey circles represent test1 ( = 45 s; MRT = 51 s) and black circles test2 ( = 41 s; MRT = 48 s) for the same subject. For clarity, exponential model and residuals were not illustrated.
test2 (CST). For all but one participant (excluded) the coefficient of regression of fitted curves was higher than 0.85, with a mean coefficient of r = 0.95 (0.89 − 0.99). Fig. 1 shows an example of a regression model (A) and an example of a rest-to-exercise transition in ST with an overlap of test1 and test2 to CST (B). The analysis of the residuals showed excellent quality models, with residues ranging randomly around the zero error, with the exception of the two excluded subjects. The amplitude was sufficiently high for adequate analysis of the response in the exercise transition. Time delay (ı) showed the greatest measurement error (46%), but in this model, where it varied freely to optimize the goodness of fit of the monoexponential regression, it has no physiological significance (it does not have the same significance as in Phase I).
4.1. V˙ O2PEAK in IST The available studies on reproducibility of V˙ O2PEAK (ST) in healthy people are widely variable as to protocol type. Thus, there are CSTs for 10 min with arm support (Jones et al., 1987), tests of two
3.4. Association between V˙ O2PEAK and on-kinetics parameters We could not find association, in a matrix of correlations, between kinetics parameters and V˙ O2PEAK (test1 and test2 ). 3.5. Reproducibility After confirmation of normal distribution for all parameters using the Shapiro–Wilk test (p > 0.05), we found that the main parameters associated with IST and CST (Tables 1 and 2 respectively) showed no significant difference between tests 1 and 2 (p > 0.05 for all paired comparisons). A qualitative analysis using the Bland–Altman plot for the three main parameters of interest (Fig. 2) was conducted to visually check the dispersion difference between tests 1 and 2 and the grand mean of these tests. The absolute difference between tests 1 and 2 and the grand mean for each transition for V˙ O2PEAK , MRT and showed no statistically significant correlation (p > 0.05 for all results), indicating that there is homoscedasticity distribution for all three parameters (Atkinson and Nevill, 1998). The coefficients of variation, LOA and other variables of reproducibility analysis are shown for all parameters in Tables 3 and 4, where we observe that the V˙ O2PEAK showed excellent reproducibility and MRT was more reproducible than . 4. Discussion This study investigated the reproducibility of an incremental step test and a 4-min constant cadence step test in providing V˙ O2PEAK and parameters of V˙ O2 on-kinetics in healthy subjects. Our results indicate that these tests provide reproducible values for these variables which were well commensurate to those described in the literature for other exercise test modalities.
Fig. 2. Bland–Altman plots for the differences (errors) against the mean for (A) V˙ O2PEAK , (B) MRT and (C) tau (). The bias line (solid) and 95% LOA (dashed) are also shown.
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
11
Table 3 Means and measures of reliability for incremental step tests to peak exercise. LOA = limits of agreements; (b) = baseline. CV = coefficient of variation. Parameters
Mean of test 1 and 2
Mean ± SD difference
95% LOA
Measurement error (%)
Systematic bias (%)
Within subject CV (%)
V˙ O2(b) (mL min−1 ) V˙ O2PEAK (mL min−1)
355
−0.1 ± 81
−159 to 160
45.0
−0.03
9.4
−383 to 260 −90 to 72 −29 to 15
13.7 42.6 13.3
2.6 5.2 4.3
3.0 7.8 3.0
Total step (n◦ .) HRpeak (bpm)
2347 190 165
62 ± 164 10 ± 5 7.1 ± 11
steps (20 cm each) with a free rate (Petrella et al., 2001), the Chester step test (Buckley et al., 2004) and a CST adjusted for age and HR, and the step height dependent on the length of the leg (Thomas et al., 1993). Because of this, comparability in healthy subjects is limited. In addition, different statistical calculations of reproducibility are available. We could not find any reproducibility studies of IST in the conditions of this study, i.e.,without support and with a step height of 20 cm in healthy people. However, a COV (V˙ O2PEAK ) of 10.4% from a modified Canadian Aerobic Fitness Test was reported (Thomas et al., 1993) in a wide age range in healthy subjects, and, coefficients of test–retest correlation of 0.97 for men and 0.98 for women were mentioned in a free cadence ST (Petrella et al., 2001). However, correlation coefficients depend on a higher or lower range of the variable studied (and sample) and detect an “association” between the test–retest variable and not necessarily reproducibility. The mean difference of 61.6 mL (2.6%) between test1 and test2 for V˙ O2PEAK is a physiologically acceptable difference for studies of aerobic power, and coupled with an accuracy of 321 mL for 95% of the values achieved in all tests and a COV of 3%, makes the determination of V˙ O2PEAK in the IST reproducible and applicable in routine aerobic evaluation. A wide study on a normal cycle ergometer found a mean absolute difference between the tests of 42 mL min−1 and a COV of 5.1% (Skinner et al., 1999), a result very similar to this study. 4.2. V˙ O2 on-kinetics in CST The kinetics of the rest-to-exercise transition for V˙ O2 is possibly influenced by the type of ergometer/instrument used for exercise, because different muscle groups and types of muscle fibers may be activated differently (Carter et al., 2000; Koga et al., 2005). The ST has peculiarities that cannot be ignored, such as intense activation of the leg eccentric muscle contraction groups, which are not usually found in the cycle ergometer exercise. The mean fundamental or Phase II of 36.8 s found in our study was similar to the 34.0 s (Özyener et al., 2001), but slower than the 21.6 s (Carter et al., 2000) or 23.9 s (Burnley et al., 2000) found in healthy untrained subjects in cycle ergometry, with a comparable Phase II amplitude (∼1500 mL) and individuals’ characteristics. The differences may be due to the different degrees of physical fitness of the individuals and different mathematical models used, in addition to reasons related to the type of exercise and ergometer. Comparative studies are needed to determine if there are actually differences in V˙ O2 kinetics related to the type of ergometer/instrument. A comparative study using O2 -deficit calculations
suggested marked differences in V˙ O2 on-kinetics between the cycle ergometer and ST, suggesting that these differences would be secondary to (i) type of muscle fibers recruited; (ii) static and dynamic components of muscle contraction and/or (iii) the muscle perfusion (di Prampero et al., 1989). The average MRT of 48.5 s in our study, however, is slower than the average values found in cycle ergometry in populations similar to ours: 43.9 s (Carter et al., 2000) and 44.1 s (Burnley et al., 2000), considering MRT = + ı at a high intensity level of exercise. The main difference is due to a slower Phase II and a shorter ı in our study. The slower average in our study, and in consequence the MRT, could be due to the transition which occurs from rest-to-exercise in ST, unlike most cycle ergometer studies which use a transition from unload-to-exercise. The difference is that in rest-to-exercise there is a greater and possibly more prolonged mobilization pool of blood through the venous return from the legs to the lungs during the cardiopulmonary phase (Phase I) of exercise, which would also explain the longer MRT in the knee extension exercise in comparison to the cycle ergometer (Koga et al., 2005), therefore including some Phase I data in the fitting model (when excluding the first 20 s of data) and, in consequence, reducing ı in such free modeling for ı, unlike ı constrained to Phase I–Phase II transitions (ı = Phase I) as in Carter et al. (2000) and Burnley et al. (2000). The fact that there are no previous references to this modality of exercise was a challenge and limited our ability to compare the reproducibility of V˙ O2 kinetics in ST. Our COV of 6.9% for Phase II and 4.5% for the MRT indicates an acceptable range for application as an exercise test and is within the range of physiological measures in humans and compares, although using different methodologies, to the 11.5% (Markovitz et al., 2004) and 11.0% (Özyener et al., 2001) for TC in untrained subjects on cycle ergometry. Similar to V˙ O2PEAK , there was no significant difference between the means of the two tests (test1 and test2 ), indicating that there was no significant bias due to the effect of learning on the second test, interfering in the results. The accuracy of 10 s for the of Phase II, suggests that for 95% of the time that the subject is tested, it is possible for an individual with, for example, a = 41 s (Fig. 1B) to obtain a new of between 31 s and 51 s. However, as suggested in Fig. 2C, 10 of the 11 subjects tested had an accuracy of 7.0 s. For the MRT, the accuracy was 9s (Table 4), therefore higher than for the Phase II and with a lower COV (4.5%). This finding was also observed in a study of reproducibility in trained individuals, where the MRT was generally more reproducible than Phase II (Kilding et al., 2005) on the treadmill.
Table 4 Means and measures of reliability (on-kinetics) for constant step tests. A = amplitude; ı = time delay; = time constant; MRT = mean response time; HR = heart rate. Parameters V˙ O2(b) (mL min−1 ) V˙ O2 ,(plateau) (mL min−1 ) −1
A (mL min ) MRT(0-3 min) (s) ı (s) (20 s-3 min) (s) HR(4 min) (bpm)
Mean of Test 1 and 2 355 1983 1627 48.5 11.7 36.8 143
Mean ± SD difference
95% LOA
Measurement error (%)
Systematic bias (%)
−39 ± 63
−84 to 164
35.0
−11.1
8.3
−23 17 −0.2 0.1 −0.3 0.4
−249 to 296 −336 to 301 −8.7 to 9.1 −5.5 to 5.4 −9.9 to 10.5 −21 to 21
13.7 19.6 18.3 46.6 27.7 14.5
−1.2 1.1 −0.4 0.9 −0.8 0.3
3.0 4.3 4.5 11.5 6.9 3.4
± ± ± ± ± ±
132.7 154.8 4.3 2.6 5.0 10
Within subject CV (%)
12
P.d.T. Müller et al. / Respiratory Physiology & Neurobiology 207 (2015) 7–13
4.3. Methodological considerations An important point to be considered is the chosen protocol for the transition rest-to-exercise and kinetics analysis, i.e., 80–90% of the maximal speed (step min−1 ) reached in IST, to optimize the signal associated with the beginning of exercise. The objective was to achieve a range of V˙ O2 high enough for a more accurate regression analysis, since a larger amplitude is directly associated with a better regression coefficient (Bell et al., 2001). However, transitions to a sub anaerobic threshold speed (moderate exercise) may be more comfortable for patients with a more limited cardiopulmonary reserve, bearing in mind that there may be a greater noise associated with a lower amplitude, plus a lower V˙ O2 signal-to-noise relation when the rest (as in ST) instead of a “baseline” load is used, as has already been noted in the COV of up to 37% in resting V˙ O2 (Markovitz et al., 2004). This limitation could be partially overcome by a low speed adopted as a “baseline” for the ST. In the incremental test, the adoption of a step of fixed height also imposes certain limitations, since the ideal would be to adjust the height of the step to the subjects’ body height (Hansen et al., 2011). Without this, the result could be very large differences in exercise time, because the load theoretically applied on ST depends on the height of the step (see Eq. (2)). Calculating the maximum speed (step min−1 ) in a difficult situation like this, could result in an exercise that is too long, with a risk of muscle injury or, alternatively, a very short exercise time (e.g.,
