Eur J Appl Physiol (2009) 107:1–9 DOI 10.1007/s00421-009-1093-7
ORIGINAL ARTICLE
Prediction of maximal oxygen uptake from submaximal ratings of perceived exertion and heart rate during a continuous exercise test: the efficacy of RPE 13 Danielle M. Lambrick Æ James A. Faulkner Æ Ann V. Rowlands Æ Roger G. Eston
Accepted: 13 May 2009 / Published online: 2 June 2009 Ó Springer-Verlag 2009
Abstract This study assessed the utility of a single, continuous exercise protocol in facilitating accurate estimates _ 2 max) from submaximal of maximal oxygen uptake (VO heart rate (HR) and the ratings of perceived exertion (RPE) in healthy, low-fit women, during cycle ergometry. Eleven women estimated their RPE during a continuous test _ 2 max). (1 W 4 s-1) to volitional exhaustion (measured VO Individual gaseous exchange thresholds (GETs) were determined retrospectively. The RPE and HR values prior to and including an RPE 13 and GET were extrapolated against corresponding oxygen uptake to a theoretical maximal RPE (20) and peak RPE (19), and age-predicted _ 2 max. There were no HRmax, respectively, to predict VO significant differences (P [ 0.05) between measured _ 2 max from (30.9 ± 6.5 ml kg-1 min-1) and predicted VO all six methods. Limits of agreement were narrowest and intraclass correlations were highest for predictions of _ 2 max from an RPE 13 to peak RPE (19). Prediction of VO _ 2 max from a regression equation using submaximal HR VO and work rate at an RPE 13 was also not significantly dif_ 2 max (R2 = 0.78, SEE = 3.42 ml kg-1 ferent to actual VO -1 _ 2 max may be min , P [ 0.05). Accurate predictions of VO obtained from a single, continuous, estimation exercise test to a moderate intensity (RPE 13) in low-fit women, particularly when extrapolated to peak terminal RPE (RPE19). The RPE is a valuable tool that can be easily employed as an adjunct to HR, and provides supplementary clinical information that is superior to using HR alone.
D. M. Lambrick (&) J. A. Faulkner A. V. Rowlands R. G. Eston School of Sport and Health Sciences, St Luke’s Campus, University of Exeter, Heavitree Road, EX1 2LU Exeter, UK e-mail:
[email protected]
Keywords RPE Maximal oxygen uptake GET Heart rate Cycle ergometry
Introduction The Borg (1998) 6-20 ratings of perceived exertion (RPE) scale is a reliable measure used to quantify, monitor and assess an individual’s exercise tolerance and level of exertion (ACSM 2008). The strong relationship between the ratings of perceived exertion and various physiological _ 2 ), heart rate (HR)] and physical [work [oxygen uptake (VO rate (WR)] markers of exercise intensity are often demonstrated during passive estimation and perceptually regulated (active production) procedures. HR is often utilised by clinicians as an assessment tool within medical evaluations prior to beginning an exercise programme, or as a means of prescribing appropriate exercise intensities and monitoring progress during a period of training (ACSM 2008). However, several factors (environmental, behavioural and dietary) can influence the submaximal HR response during exercise. The RPE have been advocated as an adjunct to HR in obtaining useful clinical information during exercise testing procedures (ACSM 2008). Yet in practice, many still do not recognise the benefits of employing the RPE and so continue to rely on HR as the principal determinant of an individual’s level of fitness and capacity for exercise. Direct measurement of maximal oxygen uptake _ 2 max) is the gold-standard method of assessing an (VO individual’s cardiorespiratory fitness and maximal capacity for exercise (Laukkanen et al. 2001). However, the highintensity nature of exhaustive stress tests may be inappropriate for certain low-fit or patient groups. Submaximal exercise testing may provide a safe, practical means of
123
2
determining the aerobic fitness of individuals who are likely to be limited by pain and fatigue, abnormal gait, or impaired balance (Noonan and Dean 2000). Recent research by Eston and colleagues (Eston et al. 2005, 2006, 2008; Faulkner et al. 2007) has focussed on identifying whether a perceptually regulated graded exercise test may be utilised to obtain an accurate prediction of _ 2 max, and provide a means of prescribing exercise VO intensity across a range of production levels. There were no statistically significant differences between measured _ 2 max and VO _ 2 max predicted from the submaximal, VO perceptually regulated exercise test in each of these aforementioned studies, regardless of gender and fitness _ 2 max was (high- and low-fit). In each of these studies, VO _ 2 values estimated by extrapolating the submaximal VO from RPE ranges of either 9–13, 9–15, 9–17 or 11–17 to RPE 19 (peak RPE value frequently reported at termination of a graded exercise test) and RPE 20 (theoretical maximal RPE). An RPE 19 is often employed as an extrapolation end-point similar to RPE 20, as previous research has demonstrated that the theoretical maximal RPE is infrequently reported at volitional exhaustion by adults (Eston et al. 2007; St Clair Gibson et al. 1999) and children alike (Barkley and Roemmich 2008; Eston et al. 2009). Unsurprisingly, the most accurate predictions of _ 2 max are often attained following extrapolation of VO _ 2 values prior to and including an RPE 17 submaximal VO (Davies et al. 2008; Eston et al. 2005, 2006, 2008; Faulkner and Eston 2007; Faulkner et al. 2007). During perceptually _ 2 max regulated procedures, acceptable estimates of VO may be obtained from lower perceptual ranges (i.e. up to an RPE 13 or 15) following a period of familiarisation (Faulkner et al. 2007). However, for low-fit individuals, the _ 2 max from a benefit of obtaining accurate estimates of VO single-bout of exercise and from a lower perceptual range is evident. A single estimation test, which is continuous in nature and coupled with the exclusion of an RPE 17 may permit a reduction in test duration (in comparison to a perceptually regulated protocol), which may have implications for the cost and time required to conduct a stress test, and would also necessitate less of an exertional and motivational effort by the exerciser. In this regard, Faulkner et al. (2009) observed that accurate estimates of _ 2 max may be obtained from a single estimation trial, up VO to and including an RPE 13 in healthy, low-fit men and women. In comparison to an estimation protocol, narrower limits _ 2 max of agreement between measured and predicted VO from an RPE 13 were observed from a perceptually regulated graded exercise test, with low-fit men and women (Faulkner and Eston 2007; Faulkner et al. 2009). In both studies, participants experienced either a 30- or 40-W incremental increase in exercise intensity every 3 min
123
Eur J Appl Physiol (2009) 107:1–9
throughout a graded exercise test to volitional exhaustion (estimation), in which they reported their perception of exertion in the final 30 s of each exercise bout. It is plausible to assume therefore that their perception of exertion may have differed throughout each 3-min period. In this regard, it is feasible to consider that a ramp protocol, whereby workload increases at a continual rate until the _ 2 max, may enable the rate of change in attainment of VO the RPE to be more closely monitored. Thus, more accurate _ 2 max and narrower limits of agreement predictions of VO than those observed in previous studies (Faulkner and Eston 2007; Faulkner et al. 2009) may potentially be obtained. It would be reasonable to assume that due to the continuous nature of a ramp protocol an individual may attain an equivalent intensity to an RPE 13 in a relatively shorter period of time than during a 3-min step test. In a recent study by Faulkner et al. (2009), low-fit men and women _ 2 max when chose to exercise between 57 and 69% of VO requested to perform a single exercise session up to an RPE 13. However, the gas exchange threshold (GET) is often used as a non-invasive indicator of an individual’s lactate threshold, and typically occurs at approximately 45–60% _ 2 max (Jones and Poole 2005). of an individual’s VO Accordingly, it is feasible to assume that the GET occurs at a comparatively lower exercise intensity (during a shorter duration test) than at RPE 13. Therefore, it is of interest to _ 2 max may be investigate whether accurate estimates of VO obtained from the GET. The purpose of this study was to assess whether a ramp _ 2 max protocol may facilitate accurate predictions of VO from an estimation procedure with low-fit women, during a single bout of moderate exercise on a cycle ergometer. It was hypothesised that a single ramp protocol would enable _ 2 max and narrower limits of accurate predictions of VO agreement than those generated from the step protocols used in previous research (Faulkner and Eston 2007; Faulkner et al. 2009). It was further hypothesised that _ 2 max would be obtained from accurate predictions of VO _ 2 corresponding to both an RPE 13 and submaximal VO GET, and that the RPE would provide predictions of _ 2 max that were similar to or more accurate than those VO obtained from age-predicted maximal HR. Moreover, it was hypothesised that GET would occur at relatively lower exercise intensity than would be experienced at RPE 13.
Method Participants Eleven healthy women (mean ± SD; age: 31.5 ± 10.8 years; height: 1.65 ± 0.06 m; body mass: 61.3 ± 8.1 kg)
Eur J Appl Physiol (2009) 107:1–9
volunteered to take part in this study. Participants were initially classified as being of ‘low fitness’ according to self-reported activity status (performing less than 1 h of physical activity per week). This was confirmed retro_ 2 max values to the spectively by comparing individual VO Shvartz and Reibold (1990) age- and gender-specific aerobic fitness norms. All participants were asymptomatic of illness or disease and free from any acute or chronic injury, as established by the ACSM (2008) participant activity readiness questionnaire (PAR-Q). Similarly, they were all classified as being of ‘low-risk’ for undertaking a maximal test to exhaustion according to the ACSM’s risk stratification procedures (ACSM 2008; body mass index: 22.4 ± 3.3 kg m-2; percent fat (bioelectrical impedance analysis): 26.8 ± 7.2%; total cholesterol: 4.47 ± 0.90 mmol-1; nonfasting blood glucose: 4.91 ± 0.81 mmol-1; systolic blood pressure: 122.7 ± 7.6 mmHg; diastolic blood pressure: 78.5 ± 5.2 mmHg). Each participant provided written informed consent prior to participation. Research was conducted in agreement with the guidelines and testing policies of the Ethics Committee of the School of Sport and Health Sciences at the University of Exeter. Procedures Each participant performed a single laboratory-based exercise test on an electromagnetically braked cycle ergometer (Lode Excallibur Sport, V2, Lode BV, Groningen, The _ 2 max. The resistance on the Netherlands) to establish VO cycle ergometer was manipulated using the Lode Workload Programmer, accurate to ±1 W, which was independent of pedal cadence. Nevertheless, participants maintained a constant cadence of 60 rev min-1 for the duration of the exercise test. On-line respiratory gas analysis occurred every 10 s throughout each exercise test via an automatic gas calibrator system (Cortex Metalyzer II, Biophysik, Leipzig, Germany). Expired air was collected continuously using a facemask. The system was calibrated prior to each test in accordance with manufacturer’s guidelines using a 3-L syringe for volume calibration and an ambient air measure for gas calibration (Gas 1 O2: 20.93; CO2: 0.03 vol%; Gas 2 O2: 14.98; CO2: 5.04 vol%). HR was recorded continuously using a wireless chest strap telemetry system (Polar Electro _ 2 , HR, T31, Kempele, Finland). All physiological [VO respiratory exchange ratio (RER), etc.] and physical (WR) markers of exercise intensity were concealed from the participants at all times. Measures Participants reported their overall perception of exertion during the exercise test using the Borg 6-20 RPE scale (Borg 1998). The 15-point scale incorporates nine verbal
3
descriptors ranging from ‘no exertion at all’ (RPE 6) to ‘maximal exertion’ (RPE 20). A rating of 6 corresponds to the level of exertion experienced during quiet seated rest, whilst a rating of 19 approximates maximal or near-maximal physical exertion. Participants had no prior experience of perceptual scaling. Prior to the exercise test, participants were familiarised with the RPE scale and received standardised instructions that provided specific commands on how to implement the scale. The RPE scale was in full view of each participant during the exercise test. Health assessment Prior to the initial exercise test and in accordance with the ACSM risk stratification procedures for determining healthy but low-fit participants, basic anthropometric (height and body mass, SECA, Hamburg, Germany) and health measurements, including body fat (Tanita, Body Composition Analyzer, BF-350, Tokyo, Japan), blood pressure (Accoson, London, UK), cholesterol and non-fasting blood glucose (Cardiochek P.A. & PTS PANELS test strips, Polymer Technology Systems Inc., Indianapolis, USA) were performed. Also, recent smoking history (current smoker or those who have quit within the last 6 months) and family history of myocardial infarction, coronary revascularisation, or sudden death in a first degree relative under the age of 65 years was assessed. If participants had two or more risk factors, as specified by the ACSM (2008), they were excluded from the study. _ 2 max Graded exercise test to establish VO The graded exercise test utilised a ramp protocol. This commenced with a 2-min baseline measurement at 0 W, after which the WR increased at a rate of 1 W every 4 s _ 2 max. Termination of the exercise (15 W min-1) until VO test was determined by a peak or plateau in oxygen con_ 2 max), the attainment of a HR within sumption (VO ±10 b min-1 of the conventional age-predicted maximum, an RER value that was equal to or exceeded 1.15, a failure to maintain the required pedal cadence or if the participant reported volitional exhaustion (Cooke 2009; Faria et al. 2005). Participants reported their perception of exertion (RPE) every 2 min for the duration of the exercise test. Determination of the gas exchange threshold Individual gas exchange thresholds (GET) were subse_ 2 max data using a cluster quently determined from the VO _ of techniques. These included the V-slope method (Beaver et al. 1986) which detects the onset of excess carbon dioxide (CO2) production in response to lactate accumu_ 2 :VCO _ 2 plot as a lation, and is denoted on the VO
123
4
_ 2 relative to the VO _ 2. disproportionate increase in VCO _ 2) Similarly, ventilatory equivalents for oxygen (V_E =VO _ 2 ) were utilised with the GET and carbon dioxide (V_E =VCO _ 2 begins to rise whilst signified as the point at which V_E =VO _ _ VE =VCO2 is maintained or decreasing. Once the GET had been identified, all data plots were assessed by trained and experienced investigators to ensure rigidity in our interpretation. Relative physical, physiological and perceptual data corresponding to the GET were then utilised to obtain _ 2 max. predictions of VO
Eur J Appl Physiol (2009) 107:1–9
distribution (Thomas and Nelson 1996), and in order to obtain an average correlation coefficient for the whole sample. A one sample t test was used to assess the difference between an RPE 13 and the average RPE reported at GET. Paired-samples t tests were also employed to assess mean _ 2 , HR differences in physical (WR) and physiological (VO and RER) variables corresponding to both an RPE 13 and GET. _ 2 max Comparison of measured and predicted VO
Data analysis _ 2 max from RPE 13 Prediction of VO Where an RPE of 13 (according to the Borg Scale) had not been reported during the test, linear regression analysis was _ 2 used to determine the corresponding submaximal VO 1 _ 2 = a ? b (RPE13 )). Individual linvalue at RPE 13 (VO _ 2 values ear regression analyses using the submaximal VO elicited prior to and including an RPE 13 were extrapolated to an RPE of 19 (RPE19; average peak RPE value reported at the termination of the exercise test), and 20 (RPE20; theoretical maximal RPE) to obtain predictions of _ 2 max. VO _ 2 max from GET Prediction of VO Firstly, linear regression analysis was employed on the RPE data in relation to WR to establish RPE at the GET (RPE at GET = a ? b (WRGET)). The RPE provided prior to and at GET were then extrapolated against the corre_ 2 data, to an RPE of 19 (GET19) and an RPE sponding VO _ 2 max. of 20 (GET20) to obtain further predictions of VO _ 2 max from HR at RPE 13 and GET Prediction of VO The corresponding HR recorded prior to and including an RPE 13, and prior to and including GET were extrapolated _ 2 to predicted HRmax according to against submaximal VO the predictive equation: 206.9 - (0.67 9 age) as stated by Gellish et al. (2007), to obtain two further predictions of _ 2 max. VO Statistical analysis All individual correlation coefficients calculated via the linear regression analyses were converted to Fisher Zr values to approximate the normality of the sampling
1
Hereafter; a subscripted number denotes the RPE value to which submaximal variables are extrapolated via linear regression analysis.
123
One-factor repeated measures analysis of variance were _ 2 max from an RPE used to compare the predictions of VO _ 13 and GET with measured VO2 max, and predictions of _ 2 max from submaximal HR equating to an RPE 13 and VO _ 2 max. If data violated assumptions of GET to measured VO sphericity using the Mauchly test, the Greenhouse–Geisser epsilon (^e) correction factor was applied to improve the validity of the F ratio. Alpha was set at 0.05 and adjusted accordingly. Intraclass correlation coefficients were calculated via the two-way mixed effects model to quantify the reproduc_ 2 max. The 95% ibility between measured and predicted VO confidence intervals were similarly calculated to attain an interval estimate of the reproducibility within a given population. The agreement (bias and random error) _ 2 max was quantified between measured and predicted VO using the 95% limits of agreement analysis (Bland and Altman 1986). Hierarchical regression analysis _ 2 max Hierarchical regression analysis, using measured VO -1 (in L min ) as the dependent variable (DV), was used to _ 2 max from assess the additive contribution of predicted VO submaximal HR (at an RPE 13 and GET; independent _ 2 max ascertained by the variable: IV) on the predicted VO RPE (RPE19, RPE20; GET19, GET20; IV). Body mass (kg) was entered as the first IV to control for body size. By assessing the additive contribution of the RPE on the _ 2 max from submaximal relationship between predicted VO HR (at an RPE 13 and GET), regression analysis allowed us to determine which variable (HR or RPE) accounted for the greatest variance, and as such the greatest additive _ 2 max. contribution to the prediction of VO _ 2 is With the consideration that the measurement of VO not always practicable, hierarchical regression analysis was additionally used to identify whether a collective model using age-predicted maximal HR (HRmaxpred) and WR elicited at an RPE 13 may provide an appropriate regres_ 2 max (ml kg-1 min-1) = sion equation, following: VO b(HRmaxpred) ? b(WR at RPE 13) ? c; to predict the
Eur J Appl Physiol (2009) 107:1–9
5
40 -1
_ 2 max, HRmax, RER) and perceptual All physiological (VO (RPE) responses elicited at the termination of the exercise test are shown in Table 1. Individual correlation coefficients calculated using the _ 2 data prior to and including an RPE 13, submaximal VO and prior to and including the GET produced average Fisher’s Zr values of 2.034 and 2.184, respectively. This equated to average (transformed) correlation coefficients of r = 0.97 and r = 0.98, for RPE 13 and GET, respectively. Similarly, average Fisher’s Zr values of 2.233 and 2.287, equating to correlations of r = 0.98, were calculated _ 2 and HR at RPE 13 and GET, respectively. An between VO average Fisher Zr value of 2.040 was also produced using _ 2 individual correlation coefficients between measured VO and RPE responses recorded throughout the exhaustive _ 2 max/RPEpeak). This revealed an exercise test (i.e. to VO _ 2 average correlation coefficient of r = 0.97 between VO and RPE to maximal exertion. A one-sample t test revealed no significant difference between an RPE 13 and the average RPE reported at GET (12.1 ± 1.8; t(10) = -1.601, P [ 0.05). Paired t tests, conducted to assess the difference between the physiolog_ 2 , HR and RER) and average WR ical responses (VO attained at an RPE 13 to GET, revealed no significant _ 2 (t(10) = 1.608, P [ 0.05) between RPE differences in VO 13 (19.4 ± 4.3 ml kg-1 min-1) and GET (18.0 ± 4.8 ml kg-1 min-1). Similarly, no significant differences (t(10) = 1.699, P [ 0.05) were observed between HR at RPE 13 (140.4 ± 15.0 b min-1) and HR at GET (134.7 ± 18.3 b min-1). The RER between RPE 13 (1.00 ± 0.06) and GET (0.98 ± 0.06) approached significance (t(10) = 2.208, P [ 0.05). Furthermore, when participants reported an RPE 13, they were exercising at a significantly higher (t(10) = 3.235, P \ 0.01) average WR (99.3 ± 18.9 W) than at GET (80.5 ± 17.3 W).
One-factor repeated-measures ANOVA demonstrated no significant differences (F(1.2,11.7) = 0.440, P [ 0.05, _ 2 max (30.9 ± 6.5 ml ^e = 0.582) between measured VO -1 -1 _ kg min ) and predicted VO2 max from RPE19 (29.6 ± 6.8 ml kg-1 min-1) or from GET19 (31.4 ± 7.7 ml kg-1 min-1). Similarly, no significant differences (F(1.3,12.7) = 0.779, P [ 0.05, ^e = 0.633) were revealed between _ 2 max from RPE20 (31.0 ± measured and predicted VO -1 -1 7.5 ml kg min ) or from GET20 (33.2 ± 8.3 ml kg-1 min-1, see Fig. 1). In addition, no significant differences (F(1.3,13.3) = 0.018, P [ 0.05, ^e = 0.662) were _ 2 max from observed between measured and predicted VO submaximal HR equating to an RPE 13 (31.0 ± 6.2 ml kg-1 min-1), or GET (30.7 ± 8.5 ml kg-1 min-1), respectively. Intraclass correlation coefficients (R) and 95% limits of _ 2 max were agreement between measured and predicted VO much higher and narrower, respectively, for the predictions _ 2 max from an RPE 13 than from either the GET or of VO submaximal HR, as shown in Table 2.
35
-1
Results
_ 2 max Analysis of measured and predicted VO
Oxygen Uptake (ml·kg ·min )
_ 2 max of the low-fit women in this study from a subVO maximal exercise test. A paired-sample t test was subsequently conducted to compare any differences between _ 2 max and the VO _ 2 max predicted using the measured VO regression equation.
30 25 20 15 10 5 0 Measured
RPE19
RPE20
GET19
GET20
VO2max
HR at RPE 13
HR at
Regression
GET
Equation
_ 2 max and predicted Fig. 1 Comparison between measured VO _ 2 elicited prior to and including an _ 2 max from submaximal VO VO RPE 13 and gas exchange threshold extrapolated to both an RPE of 19 (RPE19 and GET19, respectively) and an RPE of 20 (RPE20 and GET20, respectively); submaximal heart rate corresponding to an RPE 13 or GET extrapolated to age-predicted heart rate maximum (Gellish et al. 2007), respectively; the regression equation devised using agepredicted heart rate maximum and work rate at an RPE 13. Values are mean ± SEE
_ 2 ; heart rate, HR; respiratory exchange ratio, RER) and perceptual (ratings of perceived Table 1 Maximal physiological (oxygen uptake, VO exertion, RPE) responses elicited at termination of the exercise test (maximal work rate, WR) _ 2 max (ml kg-1 min-1) VO
HRmax (b min-1)
RERmax
Terminal RPE
WRmax (W)
30.9 ± 6.5
183 ± 7
1.16 ± 0.07
19.09 ± 0.83
176.1 ± 33.6
Values are mean ± SD
123
6
Eur J Appl Physiol (2009) 107:1–9
_ 2 max (ml kg-1 min-1) Table 2 The intraclass correlations (ICC; R) and 95% limits of agreement (LoA) between measured and predicted VO a
RPE19
a
b
b
c
HRmax (RPE 13)
c
HRmax (GET)
d
0.94
0.53
0.54
0.85
0.69
0.93
RPE20
GET19
GET20
Regression equation
ICC (R)
0.96 (0.84–0.99)
(0.76–0.98)
(-0.75 to 0.87)
(-0.71 to 88)
(0.43–0.96)
(-0.17 to 0.92)
(0.76–0.98)
95% LoA (ml kg-1 min-1)
1.28 ± 5.27
-0.41 ± 5.64
-0.50 ± 15.88
-2.31 ± 16.48
-0.1 ± 9.1
0.2 ± 14.6
0.0 ± 6.0
The confidence intervals of the ICC are reported in parentheses a
Predicted from an RPE 13
b
Predicted from the RPE equating to gas exchange threshold (GET)
c
Predicted from the heart rate at an RPE 13 and at GET, respectively
d
Predicted from the devised regression equation incorporating age-predicted heart rate maximum and work rate at an RPE 13
123
Measured VO2max (ml·kg-1·min-1)
Body mass did not account for any variance in measured _ 2 max (P [ 0.05) for any of the subsequent analyses. VO The RPE accounted for 90% (P \ 0.001) of the variance in _ 2 max (both RPE19 and RPE20) when analysed measured VO as the second IV. The additive contribution of the predic_ 2 max from HR at an RPE 13 (third IV) was a tion of VO further 7% (P \ 0.01) and 6% (P \ 0.05) when extrapolated to RPE19 and RPE20, respectively, which when combined with the RPE accounted for 97 and 96% of the _ 2 max, respectively. Predictions of total variance in VO _ 2 max from HR at an RPE 13 (second IV) did not VO significantly (18%; P [ 0.05) explain the variance in _ 2 max (DV); yet, the additive contribution of measured VO RPE19 (third IV) significantly accounted for a further 79% (P \ 0.001) and RPE20 (third IV), a further 78% _ 2 max from HR at (P \ 0.001). When predictions of VO GET and the RPE at the GET (GET19, GET20) were analysed as either the second IV or third IV, neither significantly accounted for variance (B22%; P [ 0.05) in _ 2 max. measured VO Hierarchical regression analysis revealed that HRmaxpred and WR at an RPE 13 accounted for a significant proportion (61%, P \ 0.01; and a further 17%, P \ 0.05, respectively) of the total variance (78%) in _ 2 max. Accordingly, the corresponding measured VO regression equation developed for the prediction of _ 2 max was: VO _ 2 max = 0.749 (HRmaxpred) ? 0.142 VO (WR at RPE 13) - 122.37. No significant differences (t(10) = -0.001, SEE = 3.42 ml kg-1 min-1, P [ 0.05) _ 2 max and the were observed between measured VO _ 2 max predicted from the regression equation VO (30.9 ± 5.76 ml kg-1 min-1; see Figs. 2a, b). Intraclass correlation coefficients (R) and 95% limits of agreement _ 2 max from the between measured and predicted VO regression equation are shown in Table 2.
a 40.00 35.00
30.00
25.00
20.00
15.00 20.00
25.00
30.00
35.00
40.00
Predicted VO2max (ml·kg-1·min-1)
b
10.0 8.0
LoA (ml·kg-1·min-1)
Hierarchical regression analysis
6.0 4.0 2.0 0.0 -2.0 -4.0 -6.0 -8.0 -10.0 0.0
10.0
20.0
30.0
40.0
50.0
-1
Predicted VO2max (ml·kg ·min-1 )
Fig. 2 a Scatterplot demonstrating the relationship between measured _ 2 max; ml kg-1 min-1) and predicted maximum oxygen uptake (VO _ 2 maxpred = b(HRmaxpred) ? b(WR from the regression equation: VO at RPE 13) ? c, where b is the beta value; HRmaxpred is the predicted maximum heart rate according to the equation by Gellish et al. (2007); WR at RPE 13 is the work rate attained when an RPE of 13 is first reported; c is a constant. b Limits of agreement (LoA) plot demonstrating the agreement (bias ± random error) between _ 2 max, measured and predicted maximum oxygen uptake (VO _ 2 maxpred = ml kg-1 min-1) from the regression equation: VO b(HRmaxpred) ? b(WR at RPE 13) ? c, where b is the beta value; HRmaxpred is the predicted maximum heart rate according to the equation by Gellish et al. (2007); WR at RPE 13 is the work rate attained when an RPE of 13 is first reported; c is a constant
Eur J Appl Physiol (2009) 107:1–9
Discussion This study assessed the utility of a ramp protocol in _ 2 max from submaxifacilitating accurate estimates of VO _ mal VO2 corresponding to the gaseous exchange threshold and an RPE 13 in healthy, low-fit women, during a single exercise bout on a cycle ergometer. The strong relationship between the Borg (1998) 6-20 RPE scale and physiological criteria was confirmed in this study, with a commensurate _ 2 producing a high correlation of rise in the RPE and VO r = 0.97 across the duration of the exhaustive exercise test. Strong correlation coefficients were also observed between _ 2 during the lower workloads, up to and the RPE and VO including both an RPE 13 (r = 0.97) and GET (r = 0.98). This is contrary to previous research that suggests the RPE to be more strongly correlated with physiological variables at higher intensities of exercise (Eston and Williams 1988; Faulkner et al. 2007). A strong linear relationship _ 2 and HR response to increasing (r = 0.98) between the VO exercise intensity was similarly observed prior to and including both an RPE 13 and GET. A single, low to moderate intensity estimation protocol, _ 2 max, has the which facilitates accurate predictions of VO potential to be an appropriate means of stress testing certain low-fit or clinical populations. Furthermore, when prescribing exercise to healthy low-fit populations, exercise of a lower intensity may be more appropriate as such groups are more likely to be negatively affected by the demand of high intensity exercise (Hardy and Rejeski 1989; Parfitt et al. 1996). On the basis of anecdotal evidence and past research (Faulkner et al. 2009), it was postulated that in most cases GET would occur prior to an RPE 13 being reported. As such, if accurate estimates of _ 2 max could be obtained from GET, it would have the VO advantage of further reducing the duration of a stress test. It is notable in this study that on average, GET was observed approximately 47 s prior to participants reporting an RPE 13 (5.45 and 6.32 min, respectively), and that the average workload attained at GET was statistically lower than the corresponding workload at an RPE 13. However, no statistical differences were apparent between the perceptual _ 2, responses or the physiological costs of the exercise (VO HR and RER) at GET compared to an RPE 13. Indeed for the participants in this study, GET occurred at approxi_ 2 max, in accordance with Jones mately 59% of their VO and Poole (2005), whereas at an RPE 13, participants _ 2 max, which is attained approximately 64% of their VO similar to the observations of Faulkner et al. (2007, 2009). The ACSM (2008) recommends 40–85% of oxygen _ 2 R) or 50–90% HRmax as a suitable uptake reserve (VO exercise intensity range to enhance cardiorespiratory fitness. _ 2 consumption of all participants At an RPE 13, the mean VO in the current study was 19.4 ± 4.3 ml kg-1 min-1. This
7
accords with the guidelines of the ACSM (2008) as it _ 2 R, or 5.5 METS, which is similarly in equated to 60% of VO the recommended range for moderate daily activity (3–6 METS). Comparable findings were also observed for HR, with participants attaining approximately 74 and 77% of their HRmax at GET and an RPE 13, respectively. Furthermore, the average duration of the test (6.32 min), from the onset of exercise to the participant reporting an RPE 13, was comparable to the findings of Faulkner et al. (2009). In their study, the time taken for women to report an RPE 13 _ 2 max was approximately during a graded exercise test to VO 6 min (9 min for men). Interestingly, however, predictions _ 2 max in the current study (RPE19 and RPE20) are of VO relatively more accurate than the predictions from an RPE 13 extrapolated to either an RPE19 or RPE20 in the study of Faulkner et al. (2009). Previous research (Faulkner et al. 2007) observed that _ 2 max from lower perceptual ranges (i.e. predictions of VO 9–13 or 9–15) are more accurate when extrapolated to theoretical maximal RPE (RPE20) rather than the peak terminal RPE (RPE19) consistently reported at volitional exhaustion (Eston et al. 2007; Faulkner et al. 2007; Kay et al. 2001; St Clair Gibson et al. 1999). However, this study _ 2 max demonstrated equivalently accurate estimates of VO when extrapolating from an RPE 13 to either an RPE19 or RPE20. Furthermore, no statistical differences were noted _ 2 max from GET when between measured and predicted VO extrapolated to an RPE19 or RPE20. As such, this study further demonstrates the utility of submaximal ratings of perceived exertion for providing an appropriate indication of an individual’s maximal aerobic capacity. Moreover, as there were no differences between measured and predicted _ 2 max from HR at GET or RPE 13 when extrapolated to VO age-predicted maximal HR (Gellish et al. 2007), we can also infer that HR was an accurate indicator of maximal aerobic capacity in the low-fit women in this study. Despite no apparent differences between measured and _ 2 max, measures of agreement and consistency predicted VO _ 2 max obtained from an RPE 13 indicate that estimates of VO may be more accurate than those from either the GET or submaximal HR extrapolated to age-predicted HRmax. In this regard, the predictions from an RPE 13 revealed narrower limits of agreement and higher intraclass correlation coefficients (1.28 ± 5.27 and -0.41 ± 5.64 ml kg-1 min-1; R = 0.96 and 0.94 for RPE19 and RPE20, respectively) than the predictions from the GET (-0.50 ± 15.88 and -2.31 ± 16.48 ml kg-1 min-1; R = 0.53 and 0.54 for GET19 and GET20, respectively), or HR at either RPE 13 (-0.1 ± 9.1 ml kg-1 min-1; R = 0.85) or GET (0.2 ± 14.6 ml kg-1 min-1; R = 0.69) when extrapolated to agepredicted maximal HR. It is interesting to note that when two participants are excluded from the reliability analyses, due to the
123
8
_ 2 in relation to their considerable variability in their VO RPE responses, the limits of agreement between measured _ 2 max for the remaining participants in and predicted VO this study substantially improve to 2.78 ± 6.84 and 1.14 ± 6.84 ml kg-1 min-1, and the intraclass correlation coefficients increase to R = 0.90 for both GET19 and GET20. It is plausible that the large discrepancy between predictions from GET and an RPE 13 for these two participants, despite the relative similarities in terms of physiological cost between the two conditions, may in part be the result of methodological error. The process of _ 2 values from estimates of extrapolating submaximal VO exertion via linear regression analysis is a relatively novel but reliable means of obtaining acceptable predictions of _ 2 max (Davies et al. 2008; Eston et al. 2005, 2006, 2008; VO Faulkner et al. 2007, 2009, 2008). However, the precision of the method is highly dependent upon an individuals’ ability to accurately rate his or her perception of exertion. Thus, if an individual under-estimates their perception of exertion during the lower workloads (i.e. below GET), linear extrapolation will exacerbate this error, ultimately _ 2 max. Furthermore, resulting in an over-estimation of VO as GET was identified retrospectively, a corresponding RPE value was calculated via the same linear regression analysis technique. The subsequent use of this calculated _ 2 max from the GET RPE value in the estimation of VO may, therefore, have contained inherent error and have resulted in a less accurate prediction. It is of further interest to consider the appropriateness of the limits of agreement between measured and predicted _ 2 max in this study when inferring the utility of the RPE VO during submaximal stress test procedures. For the women in this study, the identified limits of agreement mean that (in worst case scenario) a predicted value might be 6.6 ml kg-1 min-1 above or 4.0 ml kg-1 min-1 below the _ 2 up to measured value when extrapolating submaximal VO -1 and including an RPE 13 to an RPE19, or 5.2 ml kg min-1 above or 6.1 ml kg-1 min-1 below the measured value when extrapolating to an RPE20. Alternatively, this equates to a 21 or 20% margin of error between measured and pre_ 2 max (for RPE19 and RPE20, respectively). dicted VO It is interesting to note how these values compare to many recent studies in this area which have employed both perceptually regulated and estimation procedures. Faulkner et al. (2007) observed a 35% margin of error in the pre_ 2 max from an RPE range 9–13 during a diction of VO perceptually regulated graded exercise test, which is wider than that observed from the present study. It is notable that improvements in the agreement between measured and _ 2 max have been demonstrated when extrappredicted VO olating from a higher perceptual range, and following a period of familiarisation. In the original study by Eston et al. (2005), the error between measured and predicted
123
Eur J Appl Physiol (2009) 107:1–9
_ 2 max from an RPE range 9–15 reduced by 6% (from 23 VO to 17%) over the course of three perceptually regulated exercise tests. Similar improvements have also been observed in Eston et al. (2006, 2008), and Faulkner et al. (2007) following a period of familiarisation. With regards to estimation procedures, research has typically shown _ 2 max between 28 and 35% error in the prediction of VO from perceptual ratings reported prior to and including either an RPE 13 or RPE 15 (Coquart et al. 2009; Faulkner and Eston 2007; Faulkner et al. 2009). It would be prudent to acknowledge that the worst case error of 21% and 20% in the current study still represents significant variability in _ 2 max. However, when considering the estimates of VO findings from the aforementioned studies, such a margin of error may actually be considered quite favourable. The only significant distinction between the protocol adopted in the current study and that employed by Faulkner and Eston (2007) and Faulkner et al. (2009) was the application of a continuous increase in workload (ramp protocol) as opposed to an incremental step increase in workload. As such, it is feasible to consider that the continuous change in workload may have induced a more frequent appraisal of the feelings of exertion, which in turn may have facilitated _ 2 max. Furthermore, the more accurate estimates of VO continuous WR employed in the current study is in keeping with the ACSM (2008), which advocates the use of 15 W increases in WR per minute during cycle ergometry when exercise testing deconditioned individuals or patients with cardiovascular or pulmonary disease. The hierarchical regression analyses revealed that the _ 2 max from HR at an RPE 13 combined predictions of VO and from the RPE extrapolated to an RPE19 accounted for _ 2 max (97%). the greatest of the variance in measured VO Furthermore, the RPE alone accounted for 90% of the overall variance, compared to 18% explained by HR alone. Such findings are important when considering the potential clinical application of the RPE, as these results imply that clinicians may benefit highly from employing the RPE as an adjunct to HR during exercise stress testing with low-fit _ 2 may be individuals. Given that the measurement of VO unavailable or not feasible in some situations, it is plausible that the submaximal WR which equates to an RPE 13 may be used as a substitute for the corresponding submaximal _ 2 . In this regard, hierarchical regression measurement of VO analysis demonstrated that a combination of HRmaxpred and the WR at RPE 13 accounted for 78% of the overall _ 2 max. Although a corresponding regression variance in VO equation was produced from this analysis, the authors recognise that a larger sample size is necessary to enable a more rigorous and appropriate regression equation to be developed for use with a population of low-fit women. In conclusion, this study has demonstrated that accurate _ 2 max may be obtained from a single, predictions of VO
Eur J Appl Physiol (2009) 107:1–9
estimation exercise test to a moderate intensity (RPE 13) in low-fit women, particularly when extrapolated to peak terminal RPE (RPE19). Furthermore, more accurate estimates of perceived exertion and consequently more accu_ 2 max may be elicited from a rate predictions of VO continuous exercise test rather than a graded-exercise test. _ 2 max from the gas exchange threshold is Prediction of VO not recommended due to the large potential for inaccuracies. Extrapolation of submaximal HR (equating an RPE 13) to age-predicted HRmax (Gellish et al. 2007) may also _ 2 max in low-fit women. provide accurate predictions of VO Clinicians are advised that the RPE may be a valuable tool that can be easily employed as an adjunct to HR, and may provide supplementary clinical information that is superior to using HR alone. A continuous ramp protocol may reduce the duration, and thus the associated cost and motivational effort involved in an exercise stress test.
References American College of Sports Medicine (ACSM) (2008) ACSM’s guidelines for exercise testing and prescription. Lippincott Williams & Wilkins, Philadelphia Barkley JE, Roemmich JN (2008) Validity of the CALER and OMNIbike ratings of perceived exertion. Med Sci Sports Exerc 40:760–766. doi:10.1249/MSS.0b013e318161729b Beaver WL, Wasserman K, Whipp BJ (1986) A new method for detecting anaerobic threshold by gas exchange. J Appl Physiol 60:2020–2027 Bland JM, Altman DG (1986) Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1(8476):307–310 Borg G (1998) Borg’s perceived exertion and pain scales. Human Kinetics, Champaign Cooke CB (2009) Maximal oxygen uptake, economy and efficiency. In: Eston RG, Reilly T (eds) Kinanthropometry and exercise physiology laboratory manual: tests, procedures and data. Volume 2: Physiology, 3rd edn. Routledge, London, pp 174–212 Coquart J, Dubart A-E, Douillard C, Wibaux F, Garcin M (2009) Prediction of peak oxygen uptake from sub-maximal ratings of perceived exertion elicited during a graded exercise test in obese women. Psychophysiology (in press) Davies RC, Rowlands AV, Eston RG (2008) The prediction of maximal oxygen uptake from submaximal ratings of perceived exertion elicited during the multistage fitness test. Br J Sports Med 42:1006–1010. doi:10.1136/bjsm.2007.043810 Eston RG, Williams JG (1988) Reliability of ratings of perceived effort regulation of exercise intensity. Br J Sports Med 22:153– 155. doi:10.1136/bjsm.22.4.153 Eston RG, Lamb KL, Parfitt G, King N (2005) The validity of predicting maximal oxygen uptake from a perceptually-regulated graded exercise test. Eur J Appl Physiol 94:221–227. doi: 10.1007/s00421-005-1327-2 Eston RG, Faulkner JA, Mason EA, Parfitt G (2006) The validity of predicting maximal oxygen uptake from perceptually-regulated graded exercise tests of different durations. Eur J Appl Physiol 97:535–541. doi:10.1007/s00421-006-0213-x Eston RG, Faulkner JA, St Clair Gibson A, Noakes T, Parfitt G (2007) The effect of antecedent fatiguing activity on the relationship
9 between perceived exertion and physiological activity during a constant load exercise task. Psychophysiology 44:779–786. doi: 10.1111/j.1469-8986.2007.00558.x Eston R, Lambrick D, Sheppard K, Parfitt G (2008) Prediction of maximal oxygen uptake in sedentary males from a perceptuallyregulated, sub-maximal graded exercise test. J Sports Sci 26:131–139. doi:10.1080/02640410701371364 Eston RG, Lambrick DM, Rowlands AV (2009) The perceptual response to exercise of progressively increasing intensity in children aged 7-8 years: validation of a pictorial curvilinear ratings of perceived exertion scale. Psychophysiology. doi: 10.1111/j.1469-8986.2009.00826.x Faria EW, Parker DL, Faria IE (2005) The science of cycling: physiology and training—part 1. Sports Med 35:285–312. doi: 10.2165/00007256-200535040-00002 Faulkner JA, Eston RG (2007) Overall and peripheral ratings of perceived exertion during a graded exercise test to volitional exhaustion in individuals of high and low fitness. Eur J Appl Physiol 101:613–620. doi:10.1007/s00421-007-0536-2 Faulkner JA, Parfitt G, Eston RG (2007) Prediction of maximal oxygen uptake from the ratings of perceived exertion and heart rate during a perceptually-regulated sub-maximal exercise test in active and sedentary participants. Eur J Appl Physiol 101:397– 407. doi:10.1007/s00421-007-0508-6 Faulkner JA, Lambrick D, Parfitt G, Rowlands A, Eston R (2009) ˚ strandPrediction of maximal oxygen uptake from the A Ryhming nomogram and ratings of perceived exertion. In: Reilly T, Atkinson G (eds) Contemporary sport, leisure and ergonomics. Routledge, London, pp 197–214 Gellish RL, Goslin BR, Olson RE, McDonald A, Russi GD, Moudgil VK (2007) Longitudinal modelling of the relationship between age and maximal heart rate. Med Sci Sports Exerc 39:822–829. doi:10.1097/mss.0b013e31803349c6 Hardy CJ, Rejeski WJ (1989) Not what, but how one feels: the measurement of affect during exercise. J Sport Exerc Psychol 11:304–317 Jones AM, Poole DC (2005) Introduction to oxygen uptake kinetics and historical development of the discipline. In: Jones AM, Poole DC (eds) Oxygen uptake kinetics in sport, exercise and medicine, 1st edn. Routledge, London, pp 3–35 Kay D, Marino FE, Cannon J, St Clair Gibson A, Lambert MI, Noakes TD (2001) Evidence for neuromuscular fatigue during cycling in warm humid conditions. Eur J Appl Physiol 84:115– 121. doi:10.1007/s004210000340 Laukkanen JA, Lakka TA, Rauramaa R, Kuhanen R, Venalainen JM, Salonen R, Salonen JT (2001) Cardiovascular fitness as a predictor of mortality in men. Arch Intern Med 161:825–831. doi:10.1001/archinte.161.6.825 Noonan V, Dean E (2000) Submaximal exercise testing: clinical application and interpretation. Phys Ther 80:782–807 Parfitt G, Eston R, Connolly D (1996) Psychological affect at different ratings of perceived exertion in high- and low-active women: a study using a production protocol. Percept Mot Skills 82:1035–1042 Shvartz E, Reibold RC (1990) Aerobic fitness norms for males and females aged 6–75 years: a review. Aviat Space Environ Med 61:3–11 St Clair Gibson A, Lambert MI, Hawley JA, Broomhead SA, Noakes TD (1999) Measurement of maximal oxygen uptake from two different laboratory protocols in runners and squash players. Med Sci Sports Exerc 31:1226–1229. doi:10.1097/00005768199908000-00022 Thomas JR, Nelson JK (1996) Research methods in physical activity. Human Kinetics, Champaign
123
