Abstract
This study cross-validated statistical models for prediction of peak oxygen consumption using ratings of perceived exertion from the Adult OMNI Cycle Scale of Perceived Exertion. Seventy-four participants (men: n=36; women: n=38) completed a graded cycle exercise test. Ratings of perceived exertion for the overall body, legs, and chest/breathing were recorded each test stage and entered into previously developed 3-stage peak oxygen consumption prediction models. There were no significant differences (p>0.05) between measured and predicted peak oxygen consumption from ratings of perceived exertion for the overall body, legs, and chest/breathing within men (mean±standard deviation: 3.16±0.52 vs. 2.92±0.33 vs. 2.90±0.29 vs. 2.90±0.26 L·min−1) and women (2.17±0.29 vs. 2.02±0.22 vs. 2.03±0.19 vs. 2.01±0.19 L·min−1) participants. Previously developed statistical models for prediction of peak oxygen consumption based on subpeak OMNI ratings of perceived exertion responses were similar to measured peak oxygen consumption in a separate group of participants. These findings provide practical implications for the use of the original statistical models in standard health-fitness settings.
Keywords: prediction equations, cycle ergometry, undifferentiated and differentiated RPE, exercise testing
Introduction
Evaluation of peak oxygen consumption (VO2peak) using graded exercise tests aids in the development of exercise prescriptions to improve cardiorespiratory fitness. Direct assessment of VO2peak is considered the criterion measure of cardiorespiratory fitness [1], however such a laboratory-based measure is not always practical. In certain health-fitness settings where it is not appropriate for an individual to undergo peak testing conditions, subpeak exercise testing to predict VO2peak can be employed. Typically, heart rate (HR) responses are used as predictor variables in these statistical models [4, 10, 12]. However, HR for use in cardiorespiratory prediction models may be affected by many factors, in particular the error in assessment and the need for a steady state HR at each test stage. For example, determination of the beats·min−1 using radial palpation is commonly employed in standard health-fitness settings. Human error could occur such as missing a beat during palpation or extrapolating the beats·min−1 from a shorter duration of assessment (e.g., 10 or 15 seconds) rather than obtaining HR for an entire minute [16]. Additionally, the estimation of HR using wired or telemetric monitoring equipment (e.g., electrocardiography, torso transmitter devices) may not be available or accurate due to technical difficulties [15]. Predictions of VO2peak are also often based on age-predicted maximal HR equations. This could result in inaccurate VO2peak predictions, given recent evidence suggesting substantial estimation error when calculating maximal HR from an individual’s age [23, 30]. Thus, alternative methods are needed to accurately predict peak aerobic power when peak exercise testing is not practical and use of HR alone for prediction is problematic.
Ratings of perceived exertion (RPE) can be defined as a measure of the subjective intensity of effort, strain, discomfort, and/or fatigue experienced during exercise [24]. RPE are numbers from category rating scales that indicate an individual’s perceptual exertion level during physical activity [26]. The rationale for the application of RPE in practice is based on the interdependence of the physiological and perceptual responses through the full spectrum of exercise levels from rest to maximal intensity [26]. An individual’s perception of effort can be measured as either undifferentiated or differentiated RPE. According to Pandolf [25], undifferentiated RPE for the overall body (O) generally defines “…an individual's integration of various physiological sensations that have different subjective weightings.” Differentiated RPE are linked to specific cardiorespiratory and peripheral signals of effort for the chest/breathing (C) and the legs (L), respectively [27]. Measures of both undifferentiated and differentiated RPE are useful in health-fitness settings, where both global and anatomically regionalized perceptual signals are used to assess exertional tolerance and prescribe exercise intensity.
RPE have been used to predict VO2peak in a variety of patient populations employing prediction methods using data derived from load incremented exercise tests and perceptually regulated exercise protocols [3, 7–9, 19]. This application recognizes that for most test protocols RPE increases concurrently with HR and oxygen uptake as exercise intensity increases. Thus, RPE assessed during subpeak exercise tests can be used in statistical models to predict VO2peak. Recently, subpeak RPE-O, L, and C measured by the Adult OMNI Cycle Scale [28] were used to predict VO2peak in healthy college-age men and women [19]. Findings indicated the statistical models were accurate in predicting VO2peak when compared to measured VO2peak where data were derived from the same participants for which the models were developed. However, analyses involving predicted and measured VO2peak derived from independent samples is needed to determine if previously validated models can accurately translate to standard health-fitness settings.
Thus, the primary aim of the current study was to cross-validate statistical models for prediction of VO2peak using subpeak RPE derived from the Adult OMNI Cycle Scale during cycle ergometry. Secondary aims were to evaluate the accuracy of VO2peak predictions from the previously developed statistical models using only HR and combined RPE and HR (RPE/HR) with measured VO2peak. We hypothesized that VO2peak predicted from subpeak RPE-O, L, and C using previously developed statistical models would be similar to measured VO2peak in an independent cohort of participants.
Methods
Study design
A cross-sectional, single test design was employed. The protocol was approved (Approval number PRO08060266) by the local institutional review committee and all participants provided written informed consent prior to testing. Participants were instructed to refrain from alcohol consumption and strenuous physical activity 24 hours prior to testing, and to avoid food and caffeine intake as well as tobacco use 3 hours prior to testing. The study was conducted within the ethical standards of the International Journal of Sports Medicine [14].
Participants
Men and women participants were recruited using university-wide advertisements. Participants were included if they were 18–35 years of age and sedentary or recreationally active. Recreational activity was defined as participating in light to moderate intensity aerobic exercise up to 4 days a week for a total of 160 minutes.
Peak exercise test and outcome assessments
Prior to exercise testing, height (cm) was assessed with a standard stadiometer attached to a physician’s scale (Detecto, Webb City, MO). Body weight (kg) and body fat (%) was measured with a bioelectrical impedance analyzer (Tanita Corp. of America Inc., model TBF-300A, Arlington Heights, IL). Additionally, standard instructions for the Adult OMNI Cycle Scale of Perceived Exertion (Figure 1) were read to participants prior to testing as previously reported [19]. Exercise tests were performed on a cycle ergometer (Lode B.V. Medical Technology, Corival model 844, Groningen, The Netherlands) using an incremental protocol with a pedal rate of 50 revolutions·min−1 signaled by an electronic metronome (Franz Mfg. Co. Inc., model XB-700, New Haven, CT). Using the graded protocol previously reported, the test began at a power output of 75 W and increased 50 W every two minutes for men. A similar graded protocol was used for women, however, the starting power output was 50 W and increased by 25 W every two minutes. RPE for O, L, and C were measured at 90 seconds of each two minute stage in counterbalanced order. A wireless monitor (Polar Electro, Kempele, Finland) was used to measure HR (beats·min−1) immediately after the measurement of RPE. Oxygen consumption (L·min−1 and mL·kg−1·min−1) measurements were obtained from 1:45 to 2:00 minutes of each exercise stage using an automated respiratory-metabolic system (Parvo Medics, TrueOne 2400, Salt Lake City, UT). Standardized verbal encouragement was given to all subjects at the end of each stage. The test was terminated if the participant could no longer maintain the pedal cadence within 10 revolutions·min−1 of the target value for 10 consecutive seconds or self-terminated the test due to exhaustion. The following physiological and perceptual criteria for achieving VO2peak was also evaluated in men and women participants: 1) a plateau in absolute (<150 ml·min−1) or relative (<2.1 ml·kg−1·min−1) VO2 with increasing work rates, 2) HR within ±10 beats·min−1 of age-predicted maximal HR (220-age), 3) a respiratory exchange ratio (VCO2·VO2−1) ≥1.15, and 4) an OMNI RPE ≥8.
Figure 1.
Adult OMNI Cycle Scale of Perceived Exertion. Used with permission from Lippincott, Williams, and Wilkins. Med Sci Sports and Exerc 2004; 36(1): 102–108 [28].
Sample size
A similar sample size employed in a prior study [19] that initially developed and validated the VO2peak statistical models was used for the current trial. Thus, a sample of 40 men and 40 women participants was a priori chosen to evaluate the accuracy of the previously developed statistical models using measured VO2peak as the criterion variable. The target sample size was increased to a maximum of 44 participants for each sex to account for potential unevaluable exercise test data (e.g., dropouts, technical error, outliers).
Statistical analyses
The simultaneous multiple linear regression equations previously developed [19] were used to predict VO2peak from subpeak RPE and HR responses for the men and women participants in the current trial. Subpeak variables were entered into the 7 separate statistical models (Table 1) that had been previously developed for each sex. A within-subjects analysis of variance (ANOVA) was performed on measured and predicted VO2peak within each sex. Significance was set a priori at α=0.05. If a significant main effect was found among measured and predicted VO2peak within each sex, post-hoc pairwise comparisons were performed using the Bonferroni adjustment. Pearson correlations were used to determine the relation between measured VO2peak and the predicted VO2peak from each statistical model. Assumptions of parametric statistics were evaluated prior to analyses. Statistical analyses were performed with IBM SPSS Statistics (v22.0. Armonk, NY).
Table 1.
Previously validated peak oxygen consumption prediction equations for sedentary and recreationally active, college-aged men and women participants.
| Sex | Subpeak Variables |
VO2peak Prediction Models | |
|---|---|---|---|
| Men | RPE | O | 3.87−(0.034·O1)+(0.012·O2)−(0.162·O3) |
| L | 3.92+(0.005·L1)−(0.009·L2)−(0.149·L3) | ||
| C | 3.70−(0.043·C1)+(0.054·C2)−(0.161·C3) | ||
| HR | 6.118−(0.007·HR1)+(0.017·HR2)−(0.029·HR3) | ||
| RPE/HR | O/HR | 5.503+(0.004·O1)+(0.039·O2)−(0.154·O3)−(0.012·HR1)+(0.023·HR2)−(0.022·HR3) | |
| L/HR | 5.704+(0.04·L1)+(0.006·L2)−(0.132·L3)−(0.008·HR1)+(0.014·HR2)−(0.019·HR3) | ||
| C/HR | 5.810−(0.068·C1)+(0.113·C2)−(0.153·C3)+(0.000·HR1)+(0.017·HR2)−(0.029·HR3) | ||
| Women | RPE | O | 2.57−(0.073·O1)+(0.128·O2)−(0.171·O3) |
| L | 2.61−(0.006·L1)+(0.086·L2)−(0.162·L3) | ||
| C | 2.51−(0.028·C1)+(0.064·C2)−(0.130·C3) | ||
| HR | 4.337+(0.001·HR1)+(0.000·HR2)−(0.015·HR3) | ||
| RPE/HR | O/HR | 3.583−(0.082·O1)+(0.069·O2)−(0.095·O3)+(0.007·HR1)−(0.003·HR2)−(0.011·HR3) | |
| L/HR | 3.740−(0.043·L1)+(0.089·L2)−(0.123·L3)+(0.008·HR1)−(0.004·HR2)−(0.010·HR3) | ||
| C/HR | 3.692−(0.052·C1)+(0.045·C2)−(0.078·C3)+(0.008·HR1)−(0.002·HR2)−(0.013·HR3) | ||
C, ratings of perceived exertion for the chest/breathing; HR, heart rate; L, ratings of perceived exertion for the legs; O, ratings of perceived exertion for the overall body; RPE, ratings of perceived exertion; 1=subpeak variables corresponding to Stage 1; 2=subpeak variables corresponding to Stage 2; 3=subpeak variables corresponding to Stage 3
Note the slash sign (/) indicates a combination of subpeak ratings of perceived exertion and heart rate predictor variables for use in the peak oxygen consumption prediction models.
Used with permission from Ammons Scientific Ltd. Percept Mot Skills 2014; 118(3): 863–881 [19].
Results
Initial screenings and outliers
Eighty-four participants (men: n=40; women: n=44) initially gave their consent to participate. Following screenings and evaluation of outliers, four men and six women participants were removed from the trial. Assumptions of parametric statistics for 74 participants (men: n=36; women: n=38) were evaluated prior to analyses. Table 2 provides baseline characteristics for participants used in the final analyses as well as those who were deemed outliers and removed from the trial. Table 3 lists physiological and RPE responses to the cycle exercise test for those participants who were used in the final analyses.
Table 2.
Baseline characteristics for participants used in the final analyses and outliers removed from final analyses.
| Final Analyses | Outliers removed | |||
|---|---|---|---|---|
|
|
|
|||
| Characteristics | Men (n=36) |
Women (n=38) |
Men (n=3) |
Women (n=3) |
| Age (yrs) | 21.67±3.09 | 21.58±3.24 | 20.00±1.00 | 24.00±1.00 |
| Height (cm) | 177.13±5.48 | 164.56±5.38 | 185.00±5.27 | 165.67±8.96 |
| Weight (kg) | 75.93±8.95 | 64.11±9.19 | 88.83±8.20 | 75.00±4.01 |
| BMI (kg·m−2) | 24.19±2.25 | 23.66±3.19 | 25.97±3.71 | 27.37±2.80 |
| Body fat (%) | 14.25±4.25 | 26.30±6.62 | 14.70±3.75 | 29.17±8.72 |
| VO2peak (L·min−1) | 3.16±0.52 | 2.17±0.29 | 5.02±0.40 | 3.08±0.16 |
| VO2peak (mL·kg−1·min−1) | 41.92±7.25 | 34.22±4.73 | 55.97±8.46 | 41.13±1.99 |
Data are presented as mean±standard deviation. BMI, body mass index; VO2peak, peak oxygen consumption.
Table 3.
Physiological and ratings of perceived exertion variables measured during the load incremented cycle exercise test.
| Sex | Variables | Stage 1 | Stage 2 | Stage 3 | Stage 4 | Stage 5 | Stage 6 | Stage 7 | Stage 8 |
|---|---|---|---|---|---|---|---|---|---|
| Men | n (%) | 36 (100.00) | 36 (100.00) | 36 (100.00) | 35 (97.22) | 19 (52.78) | 3 (8.33) | 0 (0.00) | 0 (0.00) |
| VO2 (L·min−1) | 1.20±0.18 | 1.75±0.21 | 2.32±0.30 | 2.93±0.45 | 3.36±0.46 | 3.20±0.86 | -- | -- | |
| HR (beats·min−1) | 113.53±11.72 | 135.22±13.33 | 157.33±13.87 | 174.63±12.47 | 186.53±9.16 | 191.00±4.58 | -- | -- | |
| O | 1.11±0.95 | 3.08±1.71 | 5.89±2.00 | 8.17±1.65 | 9.26±1.15 | 9.33±0.58 | -- | -- | |
| L | 1.19±0.98 | 3.67±1.96 | 6.64±1.85 | 8.77±1.59 | 9.53±0.96 | 9.67±0.58 | -- | -- | |
| C | 1.14±1.07 | 3.19±1.67 | 5.72±1.85 | 7.80±1.89 | 9.00±1.49 | 8.33±1.53 | -- | -- | |
|
| |||||||||
| Women | n (%) | 38 (100.00) | 38 (100.00) | 38 (100.00) | 37 (97.37) | 33 (86.84) | 11 (29.73) | 1 (2.63) | 1 (2.63) |
| VO2 (L·min−1) | 0.94±0.13 | 1.16±0.15 | 1.47±0.18 | 1.79±0.23 | 2.12±0.23 | 2.29±0.28 | 2.61 | 2.78 | |
| HR (beats·min−1) | 117.53±14.33 | 134.61±15.39 | 153.42±15.16 | *168.50±14.43 | 179.12±12.81 | 176.36±14.94 | 171.00 | 176.00 | |
| O | 1.03±1.08 | 2.71±1.68 | 4.79±1.96 | 6.89±1.79 | 8.58±1.46 | 9.18±0.98 | 7.00 | 9.00 | |
| L | 1.18±1.11 | 3.08±1.73 | 5.16±1.90 | 7.30±1.78 | 8.85±1.12 | 9.55±0.82 | 9.00 | 10.00 | |
| C | 1.24±1.30 | 2.92±1.75 | 5.00±1.97 | 6.97±1.77 | 8.45±1.80 | 8.82±1.33 | 7.00 | 7.00 | |
Continuous data are presented as mean±standard deviation.
Missing heart rate for 1 woman participant during Stage 4, thus data are for n=36 only.
C, ratings of perceived exertion for chest/breathing; HR, heart rate; L, ratings of perceived exertion for the legs; O, ratings of perceived exertion for the overall body; VO2, oxygen uptake.
For men, the Shapiro-Wilk test indicated that data were not normally distributed (p<0.05) for predicted VO2peak from the RPE-L and RPE-C statistical models. Data were normally distributed for measured VO2peak and all predicted VO2peak values for women participants. A natural log transformation of VO2peak data was applied to women participant values only. Distribution of VO2peak values predicted from subpeak RPE-L was normalized but not predicted values using the RPE-C (p=0.01) statistical model. Additionally, sphericity for the main effect in both men and women was not met (p<0.001). Thus, Greenhouse-Geisser adjustments were used for interpretation of the main effect prior to any post-hoc pairwise comparisons.
Primary and secondary analyses
Figure 2 presents arithmetic means and standard deviations for measured and predicted VO2peak for men. The within-subjects ANOVA based on log-transformed data for men demonstrated significant differences among measured and predicted VO2peak (F0.22, 1.22=6.46, p<0.001). For the primary analysis, post-hoc pairwise comparisons within the men participants showed no differences between measured VO2peak and VO2peak predicted from RPE-O, L, and C (p>0.05). When using subpeak HR as predictor variables in the statistical models for men, there were no differences between measured VO2peak and the predicted VO2peak from subpeak HR, combined RPE-O/HR, and RPE-C/HR (p>0.05) models. However, the VO2peak predicted from combined subpeak RPE-L/HR was lower than measured VO2peak in the men participants (p=0.036).
Figure 2. Measured and predicted peak oxygen consumption: derived from subpeak OMNI ratings of perceived exertion, heart rate, and combined OMNI ratings of perceived exertion/heart rate for men (n=36).
Data are presented as mean±standard deviation. *Significantly lower than measured peak oxygen consumption, p=0.036.
C, ratings of perceived exertion for the chest/breathing; C/HR, combined ratings of perceived exertion for chest/breathing and heart rate statistical model; HR, heart rate statistical model; L, ratings of perceived exertion for the legs statistical model; L/HR, combined ratings of perceived exertion for the legs and heart rate statistical model; O, ratings of perceived exertion for the overall body statistical model; O/HR, combined ratings of perceived exertion for the overall body and heart rate statistical model; VO2peak, peak oxygen consumption.
For women participants, the within-subject ANOVA indicated there were significant differences among the measured and predicted VO2peak (F2.57, 95.17=12.84, p<0.001). The measured VO2peak was not significantly different than predicted VO2peak for RPE statistical models (p>0.05). There were no significant differences between measured VO2peak and predicted VO2peak from the subpeak HR and the combined subpeak RPE-L/HR and RPE-C/HR models for women (p>0.05). Post-hoc pairwise comparisons indicated a higher measured VO2peak compared to predicted VO2peak when data were derived from combined subpeak RPE-O/HR (p=0.003). Results of measured and predicted VO2peak for women are depicted in Figure 3.
Figure 3. Measured and predicted peak oxygen consumption: derived from subpeak OMNI ratings of perceived exertion, heart rate, and combined OMNI ratings of perceived exertion/heart rate for women (n=38).
Data are presented as mean±standard deviation. *Significantly different than measured peak oxygen consumption, p=0.003.
C, ratings of perceived exertion for the chest/breathing; C/HR, combined ratings of perceived exertion for chest/breathing and heart rate statistical model; HR, heart rate statistical model; L, ratings of perceived exertion for the legs statistical model; L/HR, combined ratings of perceived exertion for the legs and heart rate statistical model; O, ratings of perceived exertion for the overall body statistical model; O/HR, combined ratings of perceived exertion for the overall body and heart rate statistical model; VO2peak, peak oxygen consumption.
Correlation and limits of agreement analyses and achievement of peak oxygen consumption
Correlation analyses for men participants indicated significantly moderate to strong positive relations between log-transformed measured VO2peak and predicted VO2peak from individual RPE statistical models (RPE-O: r=0.50, p=0.002; RPE-L: r=0.57, p<0.001; RPE-C: r=0.41, p=0.013). The analysis demonstrated a moderate significant relation for measured VO2peak and VO2peak predicted from the HR model (r=0.36, p=0.03). Moderate positive relations were found between measured VO2peak and combined RPE-O/HR (r=0.49, p=0.002) and RPE-C/HR (r=0.42, p=0.011) models, with a significantly strong relation found with the RPE-L/HR (r=0.56, p<0.001) statistical model. For women participants, the analysis indicated no significant relation between measured VO2peak and VO2peak predicted from RPE-O (r=0.11, p=0.512), RPE-L (r=0.20, p=0.235), or RPE-C (r=0.14, p=0.409). The statistical model that predicted VO2peak from subpeak HR demonstrated a moderate positive relation with measured VO2peak (r=0.43, p=0.008). Moderate positive relations were found between measured VO2peak and predicted VO2peak from RPE-O/HR (r=0.36, p=0.027), RPE-L/HR (r=0.41, p=0.01), and RPE-C/HR (r=0.40, p=0.014) statistical models. An analysis of the Limits of Agreement (LOA) was evaluated between predicted and measured VO2peak for both men and women and results are depicted in Table 4 [6]. Finally, all participants voluntarily terminated the cycle exercise test or demonstrated an inability to maintain pedal cadence in addition to satisfying at least one additional physiological or perceptual criteria indicative of the attainment of VO2peak. Table 5 describes the specific indicators met for men and women participants.
Table 4.
Limits of agreement analysis for measured and predicted peak oxygen consumption for men and women.
| Sex | Prediction model |
Mean of measured and predicted VO2peak (L·min−1) |
Difference of measured and predicted VO2peak (L·min−1) |
95% LOA (L·min−1) |
|---|---|---|---|---|
| Male | O | 3.04 | 0.25 | ±0.92 |
| L | 3.03 | 0.26 | ±0.88 | |
| C | 3.03 | 0.26 | ±0.96 | |
| HR | 3.11 | 0.10 | ±1.00 | |
| O and HR | 3.09 | 0.16 | ±0.92 | |
| L and HR | 3.03 | 0.27 | ±0.90 | |
| C and HR | 3.06 | 0.21 | ±1.00 | |
|
| ||||
| Female | O | 2.10 | 0.15 | ±0.70 |
| L | 2.10 | 0.14 | ±0.64 | |
| C | 2.09 | 0.16 | ±0.66 | |
| HR | 2.16 | 0.02 | ±0.56 | |
| O and HR | 2.07 | 0.21 | ±0.60 | |
| L and HR | 2.18 | −0.02 | ±0.56 | |
| C and HR | 2.11 | 0.13 | ±0.58 | |
LOA, limits of agreement; O, overall-body; L, legs; C, chest/breathing; HR, heart rate; VO2peak, peak oxygen consumption.
Note the limits of agreement represents 2 standard deviations from the mean difference between measured and predicted VO2peak.
Table 5.
Physiological and perceptual indicators for attainment of peak oxygen consumption for both men and women participants.
| Criteria | Men Met criteria, n (%) |
Women Met criteria, n (%) |
|---|---|---|
| Plateau absolute VO2 (<150 ml·min−1) | 7 (19.4) | 7 (18.4) |
| Plateau relative VO2 (<2.1 ml·kg−1·min−1) | 10 (27.8) | 6 (15.8) |
| HR (±10 beats·min−1 of age-predicted maximal HR) | 9 (25.0) | 14 (36.8) |
| RER (VCO2·VO2−1) | 31 (86.1) | 28 (73.7) |
| OMNI RPE (O, L, or C ≥8) | 36 (100.0) | 38 (100.0) |
| OMNI RPE (O, L, or C ≥9) | 33 (91.7) | 37 (97.4) |
C, chest/breathing; HR, heart rate; L, legs; O, overall-body; RER, respiratory exchange ratio; RPE, ratings of perceived exertion; VCO2, carbon dioxide production; VO2, oxygen consumption
Note blood lactate was not assessed in the current trial.
Discussion
The primary purpose of the present study was to cross-validate previously developed VO2peak prediction models using subpeak RPE-O, RPE-L, and RPE-C measured with the OMNI Perceived Exertion Scale during cycle ergometry for young adult men and women. Twelve of the 14 (85.71%) previously developed RPE and HR models accurately predicted VO2peak in an independent cohort of men and women participants, providing strong evidence for the potential use of these equations in health-fitness settings.
Several studies have evaluated the use of subpeak RPE for predicting VO2peak for participants and patients performing a graded exercise test [7, 11, 18]. The current study however cross-validated predictive applications of the Adult OMNI Cycle Scale of Perceived Exertion. This metric has gained considerable momentum as a valid tool for application in sport and clinical and health-fitness settings for both children and adults performing a variety of exercise modalities [5, 17, 20, 22, 28, 29, 31, 32]. The OMNI scale spans a narrow numerical category range of 0–10 and provides both verbal and pictorial descriptors throughout the perceptual response range to further refine an individual’s ability to rate their perceptions of effort [28, 29]. The current cross-validation study as well as previous studies that have developed RPE-based methods to predict VO2peak, provide additional verification for the use of perceived exertion in defining peak cardiorespiratory fitness levels.
The results of the present trial suggest that both subpeak RPE and HR may be valid for predictive purposes, as they are linearly related throughout graded exercise from low to maximal levels. Unexplained variance may have caused the mismatch of measured VO2peak vs. VO2peak predicted from subpeak RPE-L/HR for men and RPE-O/HR for women. However, differences in the arithmetic means of measured and predicted VO2peak were minimal for men (0.27 L·min−1) and women (0.21 L·min−1), in addition to demonstrating significant positive moderate and strong relations and a lack of wide 95% LOA when compared to the other VO2peak predictions. The inclusion criteria for level of fitness and type of physical activity regularly performed were contingent on participant self-report of previous and current physical activity patterns for both the current trial and the study that validated the VO2peak prediction models. Self-reported physical activity is said to be adversely affected by unreliable reporting or natural variations in behavior over time [2]. Thus, trials developing predictive models should provide more specific criteria or direct testing to determine fitness levels of participants, as the current equations may only be valid for a specific subset of college-age men and women. The current study participants may have underestimated the amount of exercise they typically accomplish, thus the non-significant but lower predicted VO2peak values. Our findings however agree with the conclusions of the initial validation study [19] that VO2peak statistical models using subpeak RPE or HR alone, in addition to combinations of RPE/HR, could potentially be used in standard health-fitness settings.
From a practical application perspective, the models could potentially be employed in health-fitness, sport, and selected clinical settings to estimate VO2peak. Many of these facilities may lack automated and often expensive respiratory-metabolic analyzers. Thus, the translation of laboratory-based statistical models that employ subpeak RPE to predict oxygen uptake could be useful, in particular if HR assessment is difficult to obtain or inaccurate. Once the predicted VO2peak is determined using the statistical models, the exercise prescription of clients could be refined, specifically outlining the optimal range of subpeak exercise intensity based on the predicted peak value. An important next step however, is to determine the accuracy of the models for predicting VO2peak in a standard health-fitness setting. The current trial was based in a laboratory setting to evaluate the efficacy of the prediction models, where the optimal exercise setting and standardized testing procedures were assured. Future trials should evaluate the translation of the statistical models to standard health-fitness settings to ensure their feasibility and effectiveness in healthy college-age men and women participants where variability in the assessment of outcomes could differ. Additionally, the validation of relative VO2peak prediction models for weight bearing exercise (such as treadmill walking and running) is needed to account for the range of potential modalities that individuals may use for the completion of physical activity.
An important aspect of this cross-validation study was the evaluation of VO2peak prediction models that included submaximal undifferentiated and differentiated RPE responses. With various exercise modalities, the anatomical region of the most active skeletal muscle may provide the dominant (i.e. highest) RPE throughout an exercise test [13]. This differentiated RPE response (e.g. RPE-L during cycle exercise) may provide a more accurate prediction of peak aerobic power since local muscle fatigue is a primary reason for test termination. It has been suggested that the dominant differentiated RPE signal may explain a greater amount of the variance in VO2peak prediction in comparison to undifferentiated RPE [13]. However, the results of the current investigation suggest undifferentiated and differentiated RPE responses are similar in their ability to accurately predict VO2peak. It may be of no consequence which RPE are obtained during a submaximal exercise test, as the critical requisite step is to ensure the undifferentiated or differentiated RPE are used with the appropriate statistical model.
This cross-validation trial provides strong evidence for the use of OMNI RPE to predict VO2peak. Further validation studies from other research laboratories are needed to ensure the statistical models derived from subpeak OMNI RPE are efficacious for predicting VO2peak in the participant cohort that was studied. However, the current findings demonstrate the utility of RPE for predicting VO2peak, resulting in accurate assessment of cardiorespiratory fitness when true peak exercise testing is not feasible or warranted in young adult men and women.
Acknowledgments
Preliminary data for this project were presented at the American College of Sports Medicine’s 56th Annual Meeting [21]. The current study was supported by a University of Pittsburgh School of Education Research grant. Dr. Mays is principal investigator of two National Institutes of Health grants funded by the National Heart, Lung, and Blood Institute (K01HL115534) and the Mountain West Clinical Translational Research - Infrastructure Network National Institute of General Medical Sciences (under the parent award 1U54GM104944). Dr. Mays also receives funding from Providence Medical Group for the conduct of quality improvement projects. Neither of these grants or consultant work are related to the current study.
Footnotes
disclosures
The other authors note no conflicts.
References
- 1.ACSM. ACSM's guidelines for exercise testing and prescription. 9. Baltimore, MD: Lippincott Williams and Wilkins; 2014. p. 480. [Google Scholar]
- 2.Ainsworth BE, Caspersen CJ, Matthews CE, Masse LC, Baranowski T, Zhu W. Recommendations to improve the accuracy of estimates of physical activity derived from self report. J Phys Act Health. 2012;9(Suppl 1):S76–84. doi: 10.1123/jpah.9.s1.s76. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Al-Rahamneh HQ, Eston RG. Prediction of peak oxygen consumption from the ratings of perceived exertion during a graded exercise test and ramp exercise test in able-bodied participants and paraplegic persons. Arch Phys Med Rehabil. 2011;92:277–283. doi: 10.1016/j.apmr.2010.10.017. [DOI] [PubMed] [Google Scholar]
- 4.Astrand PO, Rhyming I. A nomogram for calculation of aerobic capacity (physical fitness) from pulse rate during sub-maximal work. J Appl Physiol. 1954;7:218–221. doi: 10.1152/jappl.1954.7.2.218. [DOI] [PubMed] [Google Scholar]
- 5.Barkley JE, Roemmich JN. Validity of the CALER and OMNI-bike ratings of perceived exertion. Med Sci Sports Exerc. 2008;40:760–766. doi: 10.1249/MSS.0b013e318161729b. [DOI] [PubMed] [Google Scholar]
- 6.Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–310. [PubMed] [Google Scholar]
- 7.Coquart JB, Eston RG, Lemaitre F, Bart F, Tourny C, Grosbois JM. Prediction of peak oxygen uptake from ratings of perceived exertion during a sub-maximal cardiopulmonary exercise test in patients with chronic obstructive pulmonary disease. Eur J Appl Physiol. 2015;115:365–372. doi: 10.1007/s00421-014-3023-6. [DOI] [PubMed] [Google Scholar]
- 8.Coquart JB, Garcin M, Parfitt G, Tourny-Chollet C, Eston RG. Prediction of maximal or peak oxygen uptake from ratings of perceived exertion. Sports Med. 2014;44:563–578. doi: 10.1007/s40279-013-0139-5. [DOI] [PubMed] [Google Scholar]
- 9.Eston R, Evans H, Faulkner J, Lambrick D, Al-Rahamneh H, Parfitt G. A perceptually regulated, graded exercise test predicts peak oxygen uptake during treadmill exercise in active and sedentary participants. Eur J Appl Physiol. 2012;112:3459–3468. doi: 10.1007/s00421-012-2326-8. [DOI] [PubMed] [Google Scholar]
- 10.Fairbarn MS, Blackie SP, McElvaney NG, Wiggs BR, Pare PD, Pardy RL. Prediction of heart rate and oxygen uptake during incremental and maximal exercise in healthy adults. Chest. 1994;105:1365–1369. doi: 10.1378/chest.105.5.1365. [DOI] [PubMed] [Google Scholar]
- 11.Faulkner J, Eston R. 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. 2007;101:613–620. doi: 10.1007/s00421-007-0536-2. [DOI] [PubMed] [Google Scholar]
- 12.Golding LA, Meyers CR, Sinning WE. Y's way to physical fitness. 3. Champaign, IL: Human Kinetics; 1989. p. 192. [Google Scholar]
- 13.Haile L, Gallagher MJ, Robertson RJ. Perceived exertion laboratory manual: from standard practice to contemporary application. New York, NY: Springer Science+Business Media; 2015. p. 322. [Google Scholar]
- 14.Harriss DJ, Atkinson G. Ethical standards in sport and exercise science research: 2014 update. Int J Sports Med. 2013;34:1025–1028. doi: 10.1055/s-0033-1358756. [DOI] [PubMed] [Google Scholar]
- 15.Janz KF. Use of heart rate monitors to assess physical activity. In: Welk GJ, editor. Physical Activity Assessments for Health-Related Research. Champaign, IL: Human Kinetics Publishers, Inc.; 2002. pp. 143–161. [Google Scholar]
- 16.Kobayashi H. Effect of measurement duration on accuracy of pulse-counting. Ergonomics. 2013;56:1940–1944. doi: 10.1080/00140139.2013.840743. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Krause MP, Goss FL, Robertson RJ, Kim K, Elsangedy HM, Krinski K, da Silva SG. Concurrent validity of an OMNI rating of perceived exertion scale for bench stepping exercise. J Strength Cond Res. 2012;26:506–512. doi: 10.1519/JSC.0b013e318220df8e. [DOI] [PubMed] [Google Scholar]
- 18.Lambrick DM, Faulkner JA, Rowlands AV, Eston RG. Prediction of maximal oxygen uptake from submaximal ratings of perceived exertion and heart rate during a continuous exercise test: the efficacy of RPE 13. Eur J Appl Physiol. 2009;107:1–9. doi: 10.1007/s00421-009-1093-7. [DOI] [PubMed] [Google Scholar]
- 19.Mays RJ, Goss FL, Nagle EF, Gallagher M, Schafer MA, Kim KH, Robertson RJ. Prediction of VO2peak using OMNI Ratings of Perceived Exertion from a submaximal cycle exercise test. Percept Mot Skills. 2014;118:863–881. doi: 10.2466/27.29.PMS.118k28w7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mays RJ, Goss FL, Schafer MA, Kim KH, Nagle-Stilley EF, Robertson RJ. Validation of adult OMNI perceived exertion scales for elliptical ergometry. Percept Mot Skills. 2010;111:848–862. doi: 10.2466/05.06.PMS.111.6.848-862. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Mays RJ, Gallagher MJ, Haile L, Goss FL, Nagle EF, Robertson RJ. Cross-validation of VO2peak prediction equations using submaximal OMNI RPE responses. Med Sci Sports Exerc. 2009;41:S213–214. [Google Scholar]
- 22.Nakamura FY, Perandini LA, Okuno NM, Borges TO, Bertuzzi RC, Robertson RJ. Construct and concurrent validation of OMNI-Kayak rating of perceived exertion scale. Percept Mot Skills. 2009;108:744–758. doi: 10.2466/PMS.108.3.744-758. [DOI] [PubMed] [Google Scholar]
- 23.Nes BM, Janszky I, Wisloff U, Stoylen A, Karlsen T. Age-predicted maximal heart rate in healthy subjects: The HUNT fitness study. Scand J Med Sci Sports. 2013;23:697–704. doi: 10.1111/j.1600-0838.2012.01445.x. [DOI] [PubMed] [Google Scholar]
- 24.Noble BJ, Robertson RJ. Perceived Exertion. Champaign, IL: Human Kinetics; 1996. p. 336. [Google Scholar]
- 25.Pandolf KB. Differentiated ratings of perceived exertion during physical exercise. Med Sci Sports Exerc. 1982;14:397–405. [PubMed] [Google Scholar]
- 26.Robertson RJ. Perceived Exertion for Practitioners: Rating Effort with the OMNI Picture System. Champaign, IL: Human Kinetics; 2004. p. 184. [Google Scholar]
- 27.Robertson RJ, Gillespie RL, McCarthy J, Rose KD. Differentiated perceptions of exertion: part II. relationship to local and central physiological responses. Percept Mot Skills. 1979;49:691–697. doi: 10.2466/pms.1979.49.3.691. [DOI] [PubMed] [Google Scholar]
- 28.Robertson RJ, Goss FL, Dube J, Rutkowski J, Dupain M, Brennan C, Andreacci J. Validation of the adult OMNI scale of perceived exertion for cycle ergometer exercise. Med Sci Sports Exerc. 2004;36:102–108. doi: 10.1249/01.MSS.0000106169.35222.8B. [DOI] [PubMed] [Google Scholar]
- 29.Robertson RJ, Goss FL, Boer NF, Peoples JA, Foreman AJ, Dabayebeh IM, Millich NB, Balasekaran G, Riechman SE, Gallagher JD, Thompkins T. Children's OMNI scale of perceived exertion: mixed gender and race validation. Med Sci Sports Exerc. 2000;32:452–458. doi: 10.1097/00005768-200002000-00029. [DOI] [PubMed] [Google Scholar]
- 30.Sarzynski MA, Rankinen T, Earnest CP, Leon AS, Rao DC, Skinner JS, Bouchard C. Measured maximal heart rates compared to commonly used age-based prediction equations in the Heritage Family Study. Am J Hum Biol. 2013;25:695–701. doi: 10.1002/ajhb.22431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Utter AC, Robertson RJ, Green JM, Suminski RR, McAnulty SR, Nieman DC. Validation of the Adult OMNI Scale of perceived exertion for walking/running exercise. Med Sci Sports Exerc. 2004;36:1776–1780. doi: 10.1249/01.mss.0000142310.97274.94. [DOI] [PubMed] [Google Scholar]
- 32.Utter AC, Kang J, Nieman DC, Dumke CL, McAnulty SR. Validation of OMNI scale of perceived exertion during prolonged cycling. Med Sci Sports Exerc. 2006;38:780–786. doi: 10.1249/01.mss.0000210201.25447.3b. [DOI] [PubMed] [Google Scholar]