Abstract
Background Submaximal exercise testing can have a greater application in clinical settings because peak exercise testing is generally not available. In previous work, a prediction equation was developed to estimate peak oxygen consumption (V̇o2) using a total body recumbent stepper (TBRS) and the Young Men's Christian Association (YMCA) protocol in adults who were healthy.
Objective The purpose of the present study was to cross-validate the TBRS peak V̇o2 prediction equation in older adults.
Design A cross-sectional study was conducted.
Methods Thirty participants (22 female, 8 male; mean age=66.8 years, SD=5.52; mean weight=68.51 kg, SD=13.39) who previously completed a peak exercise test and met the inclusion criteria were invited to participate in the cross-validation study. Within 5 days of the peak V̇o2 test, participants completed the TBRS submaximal exercise test. The TBRS submaximal exercise test equation was used to estimate peak V̇o2. The variables in the equation included age, weight, sex, watts (at the end of the submaximal exercise test), and heart rate (at the end of the submaximal exercise test).
Results A strong correlation was found between the predicted peak V̇o2 and the measured peak V̇o2. The difference between the values was 0.9 mL·kg−1·min−1, which was not statistically different. The standard error of the estimate was 4.2 mL·kg−1·min−1.
Limitations The sample included individuals who volunteered to perform a peak exercise test, which may have biased the results toward those willing to exercise to fatigue.
Conclusion The data suggest the TBRS submaximal exercise test and prediction equation can be used to predict peak V̇o2 in older adults. This finding is important for health care professionals wanting to provide information to their patients or clients regarding their fitness level.
Measurements of peak or maximum oxygen consumption (V̇o2) can be obtained via indirect calorimetry using a treadmill, cycle ergometer, or recumbent stepper by progressively increasing workload (ie, graded exercise test [GXT]) until the highest level of physiologic exertion is reached. Although peak V̇o2 testing is considered the “gold standard” for assessing aerobic capacity,1 this approach may have limited application if individuals are not able to meet the criteria for maximal exertion due to fatigue, motivation,2 or physical limitations.2,3 Therefore, conducting a maximal GXT to obtain measurements of peak V̇o2 is not always practical.
The use of a submaximal exercise test to predict peak V̇o2 is considered safer than maximal exercise testing and may be more appropriate in clinical practice and commercial settings, especially when assessing older adults.2,4 Submaximal exercise tests can be delivered in a variety of settings (gym, hallway, hospital, or clinic) and with various exercise equipment such as treadmills, cycle ergometers, and total body recumbent steppers (TBRSs). Submaximal exercise tests, such as the Six-Minute Walk Test or step test, can be performed without major exercise equipment. As highlighted in the review article by Noonan and Dean2 and by the American College of Sports Medicine,1 consideration must be given to using standardized testing procedures to ensure valid results. This consideration is particularly important in clinical practice where results of the submaximal exercise test may be used to measure progress toward goals, assessing the physiologic response to cardiovascular stress, and even in discharge planning.
In our previous work, we used an existing and reliable submaximal protocol, the Young Men's Christian Association (YMCA) cycle ergometry test, and adapted it for use on a TBRS.5 We developed a peak V̇o2 prediction equation using 5 variables: age (in years), weight (in kilograms), sex, and both heart rate (HR) and workload (in watts) at the end of the submaximal exercise test. This equation was tested in individuals 18 to 60 years old and then cross-validated in another cohort.5 The TBRS is predominantly used4,6 as the “modality of choice” for older adults.7 Therefore, the purpose of the present study was to cross-validate the recently developed peak V̇o2 prediction equation in adults 60 to 80 years of age and determine whether the equation would predict peak V̇o2 in people outside of the age range for which the equation was originally developed. We hypothesized that the peak V̇o2 prediction equation would be strongly correlated (r≥.85) and would predict the measured peak V̇o2 (R2≥.75).
Method
Study Design
A cross-sectional study design with a sample of convenience was used. Institutionally approved written consent was obtained prior to enrollment.
Participants
Between December 2012 and June 2013, 30 individuals were recruited from the community to participate in this study. Cardiac risk was determined using the American College of Sports Medicine's Guidelines for Exercise Testing and Prescription.1 Cardiac risk and participant demographics are included in Table 1. Inclusion criteria were: (1) between 60 and 80 years of age, (2) completed a GXT using the Bruce protocol or modified Bruce protocol with a respiratory exchange ratio of ≥1.1 indicating maximal effort,8 and (3) ability to travel or have transportation to participate in a submaximal exercise test within 5 days of the GXT. Individuals were excluded if they were unable to use the recumbent stepper (ie, orthopedic restrictions or unable to coordinate the arm and leg motion). If an individual demonstrated absolute exercise test termination criteria1 during the submaximal exercise test whereby early test termination was necessary, the data were not included for analysis.
Participant Demographics (N=30)a
Peak Exercise Testing
Participants who completed a peak exercise test (as part of an ongoing community fitness database) and met the inclusion criteria were invited to participate in a study to conduct a submaximal exercise test to predict peak V̇o2. All peak exercise tests were performed at the University of Kansas Medical Center Research in Exercise and Cardiovascular Health (REACH) Laboratory. Our method for peak exercise testing has been reported elsewhere.5 Briefly, participants were informed not to consume food or drink (except water) within 2 to 3 hours of the exercise test and to avoid caffeinated products for 6 hours prior to the exercise test. Participants were asked to avoid vigorous physical activity for 24 hours prior to exercise testing. Measurements of height, weight, pre-exercise HR, and blood pressure (BP) were obtained prior to exercise testing. Oxygen uptake was measured and analyzed through collection of expired gases using the Parvo Medics metabolic measurement system (Parvo Medics Inc, Sandy, Utah). Gas and flowmeter calibrations were performed on the metabolic cart according to the specifications of the manufacturer. The exercise test used was either the Bruce protocol or the modified Bruce protocol, which are both widely recognized and used treadmill tests.2 The Bruce protocol is a 3-minute incremental GXT that begins at 1.7 mph and 10% grade.9 The grade increases 2% at each stage along with concomitant increases in speed. The modified Bruce protocol is also a 3-minute incremental exercise test that starts at 1.7 mph, but the grade is 0%. The next stage is a 5% grade at 1.7 mph and progresses to the original Bruce protocol at 10% grade at 1.7 mph and then follows the same protocol progression.
Submaximal Exercise Test
Individuals who completed a peak exercise test and met the inclusion criteria were invited to participate in the submaximal exercise study. Once written informed consent was obtained, participants were scheduled within 5 days of the GXT and at similar times to complete the submaximal exercise test in the REACH Laboratory. Participants who were taking medications (or took medications prior to the peak exercise test) were instructed to follow an identical routine. All participants were familiarized with the motion of the recumbent stepper, explained the procedures, and had an opportunity to use the exercise device to practice the alternating, reciprocal movement pattern and step rate prior to the day of the submaximal exercise test.
Resting HR and BP were assessed prior to testing. The day of the submaximal exercise test, participants were fitted with a coded Polar HR monitor (Polar Electro Oy, Kempele, Oulu, Finland) for continuous use during the submaximal exercise test. The YMCA protocol (Tab. 2) we used was adapted for the recumbent stepper (T5XR, NuStep Inc, Ann Arbor, Michigan) and described in our previous work.5 We used the T5XR model because it has the menu option for constant power (in watts), which standardizes the workload across participants. Individuals were instructed to maintain a step rate between 90 and 100 steps/min. Participants started the test at 30 W, and resistance was increased every 3 minutes according to the protocol (Tab. 2) until one of the following: (1) volitional fatigue,1 (2) 85% of maximum age-predicted HR ([220−age] × 0.85) was achieved, or (3) the submaximal exercise test was completed. Ten seconds prior to the end of the second and third minutes of each stage, HR and rate of perceived exertion (RPE)10 were recorded. If these 2 measurements were within 5 bpm of each other, participants progressed to the next stage.1 If the difference was greater than 5 bpm, an additional minute of testing was performed to ensure a steady state. Upon completion of the exercise test, HR, BP, and RPE were recorded, and the individual continued to step at a comfortable self-selected speed with resistance between 30 and 40 W for 2 minutes. After the cool-down, HR and BP were assessed to ensure these values returned to near-baseline levels.
Total Body Recumbent Stepper Submaximal Exercise Test and the Young Men's Christian Association (YMCA) Protocola
Demographic information and submaximal exercise testing data were used in the equation that was previously developed to predict peak V̇o2 (mL·kg−1·min−1)5:
Predicted peak V̇o2 (mL·kg−1·min−1) =125.707+(−0.476) (age)+(7.686) (sex[0=female; 1=male])+(−0.451) (weight)+(0.179) (Wend_submax)+(−0.415) (HRend_submax)
Sample Size Justification
For this cross-validation study, identical methods to our previous work for determining the sample size were used.5 Briefly, using our prediction model with the 5 predictors (age, sex, weight, workload, and HR) and a small value for shrinkage (0.10) with the original coefficient of determination (R2) of .85,5 30 participants would be needed.11 Algina and Keselman11 suggested the difference between the original coefficient of determination and the new predicted value should be small, ranging from 0.25 to 0.10. Because the prediction equation was being testing on a group of people outside of the range for which it was developed, we chose to be conservative (difference of ≤0.10).
Data Analysis
The arithmetic mean and standard deviation were used for descriptive statistics. Pearson correlation coefficients (r) were used to determine the relationship between the predicted and measured peak V̇o2. We calculated whether the amount of shrinkage (difference) between the coefficient of determination from this study and the original value (R2=.85) from our previous work5 would be within .10,11 or R2 ≥.75. Paired t tests were used to examine whether the predicted and measured peak V̇o2 values were significantly different. The standard error of the estimate (SEE) was calculated (SEE = Σ(y′ − y)2̂/N)) to estimate the accuracy of the predicted peak V̇o2 and measured peak V̇o2. To determine the level of agreement between the actual and predicted peak V̇o2, we constructed a Bland-Altman plot. Data analysis was performed with SPSS version 20.0 for Windows (SPSS Inc, Chicago, Illinois). P values less than .05 were considered significant.
Role of the Funding Source
Dr Billinger was supported, in part, by grant K01HD067318 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Ms Mattlage and Mr Sisante were supported, in part, by grant T32HD057850 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. The content is solely the responsibility of the authors and does not necessarily represent the official views of the Eunice Kennedy Shriver National Institute of Child Health and Human Development or the National Institutes of Health. Research in the exercise and cardiovascular health laboratory space was supported by the Georgia Holland Endowment Fund.
Results
Individuals who met the inclusion criteria from their peak exercise test were invited to participate in the submaximal exercise test study. All participants (N=30) were living in the community and could ambulate independently. Participants had no functional limitations that would preclude using a treadmill for performing a GXT or the recumbent stepper reciprocal motion for the submaximal exercise test. One individual was taking a beta-blocker, which is known to lower HR during exercise. No adverse events were reported during the submaximal exercise test. During the TBRS submaximal exercise test, 29 participants reached the 85% of maximum age-predicted HR criterion, and 1 participant chose to end the test due to volitional fatigue.
Values were entered into the peak V̇o2 prediction model. The coefficient of determination between the predicted and measured peak V̇o2 values was calculated. We then figured the amount of shrinkage (difference) between this coefficient of determination (R2=.76) and the value (R2=.85) from our original model.5 The cross-validation revealed shrinkage less than .10, confirming that the model was successful in predicting peak V̇o2 in 60- to 80-year-old adults ranging from zero to multiple chronic conditions. Predicted mean peak V̇o2 was 31.36±8.32 mL·kg−1·min−1, and the measured V̇o2 was 30.43±6.34 mL·kg−1·min−1, with no significant difference found between these values (t=−1.22, P=.23; 95% confidence interval of the mean difference was −0.63, 2.50). The mean difference between the predicted peak V̇o2 and measured peak V̇o2 was 0.9 mL·kg−1·min−1, and the SEE was 4.2 mL·kg−1·min−1. The Pearson correlation coefficient was strong (r=.87, P<.001; Fig. 1) between actual and predicted and mean peak V̇o2 (mL·kg−1·min−1). The Bland-Altman plot (Fig. 2) suggests that with greater peak V̇o2 values, the equation overestimates measured peak V̇o2. However, the majority of the predicted peak V̇o2 values were within 1 metabolic equivalent (1 MET=3.5 mL·kg−1·min−1).
Relationship between predicted peak oxygen consumption (V̇o2) and measured peak V̇o2.
Bland-Altman plot of predicted and observed peak oxygen consumption (V̇o2) values. MET=metabolic equivalent.
Discussion
This cross-validation study examined whether the TBRS submaximal exercise test and the peak V̇o2 prediction equation would accurately predict measured peak V̇o2 in older adults 60 to 80 years of age. This cross-validation was important to determine since the original peak V̇o2 prediction equation was developed for individuals 18 to 60 years of age. We hypothesized that the R2 value for the cross-validation of the TBRS submaximal exercise test needed to be ≥.75. Our hypothesis was supported by our findings. The cross-validation revealed shrinkage less than .10 (R2=.76), confirming that the TBRS submaximal exercise test and the previously established equation predicted V̇o2 peak in these participants. We also report a strong correlation between the actual and predicted peak V̇o2 values.
The TBRS submaximal exercise test and prediction equation used in this study to predict peak V̇o2 have important implications for fitness environments and health care professionals. Submaximal exercise tests are safer and convenient to use and require less expensive equipment than maximal exercise testing with gas analysis. As recommended by Noonan and Dean,2 we used standardized procedures for instructing the participant on activity, food, and caffeine intake prior to the TBRS submaximal exercise test. All participants had the opportunity to practice the reciprocal motion and step rate on the TBRS prior to the submaximal test. We did not provide any information or verbal encouragement on their performance or protocol “track” selection.
In addition, we chose to conduct the submaximal exercise test on the NuStep T5XR because of the constant power option. This option allows the device to maintain a constant power output and flexibility in the step rate, similar to the cycle ergometer, which are crucial for standardizing the workload across all participants. Rehabilitation specialists such as physical therapists and fitness professionals implementing these standardized procedures with access to the NuStep T5XR can use the TBRS submaximal exercise test and previously established prediction equation to estimate peak V̇o2 in their older adult clients or patients. This protocol will allow for assessment of submaximal exercise performance, more accurate exercise prescription, and possibly improved outcomes. However, this submaximal exercise test protocol has not been tested on other NuStep devices (TRS 4000). To our knowledge, the TBRS submaximal exercise test has not been studied to determine test-retest reliability. Furthermore, it will be important to study whether the TBRS submaximal exercise test is sensitive to detect change in fitness levels and warrants further study.
As mentioned in our prior work3,5 and by other authors,4,6,7 the TBRS design is comfortable7 and has the ability to accommodate a variety of physical deficits such as impaired balance, muscle weakness from stroke, and incoordination.12 These advantages allow health care professionals or fitness trainers the ability to use this device with a variety of people with multiple chronic conditions. As shown in Table 1, our participants had a variety of chronic conditions and cardiac risk. These characteristics may allow the findings to be more generalizable to community-living individuals seeking participation in exercise. The participants in our original study had low to moderate cardiac risk and were 18 to 60 years of age.5 For this cross-validation study, we wanted to test whether the peak V̇o2 prediction equation would accurately predict peak V̇o2 in older adults (ie, 60–80 years of age). Although our goal was to recruit individuals in this age range, our mean age was 66.8 years. As shown in Table 1, the majority of the participants who volunteered for the study were 60 to 70 years of age, which may limit interpretation of our findings for those 70 to 80 years of age.
The mean difference between predicted peak V̇o2 (31.36±8.32 mL·kg−1·min−1) and actual peak V̇o2 (30.43±6.34 mL·kg−1·min−1) was less than 1 mL·kg−1·min−1 and was not significantly different. The SEE measure reflects the accuracy of the predicted peak V̇o2 measure. A high SEE value would indicate inaccuracy of the submaximal exercise test or the prediction equation, or both.13 In this study, the SEE was 4.2 mL·kg−1·min−1. Although we would have liked to have the SEE within 1 MET (3.5 mL·kg−1·min−1), a SEE value of less than 5 mL·kg−1·min−1 is considered appropriate for predicting peak V̇o2.13,14 Furthermore, the 95% limits of agreement should not be greater than ±10 mL·kg−1·min−1.14 As illustrated in Figure 2, the upper limit is 9 mL·kg−1·min−1, and the lower limit is 7 mL·kg−1·min−1. Because the 95% limits of agreement are below the recommended 10 mL·kg−1·min−1, the prediction equation has better predictive capabilities than tests that exceed the recommended value.
The Bland-Altman plot shows that the majority of our sample fell within 1 MET, and 2 individuals exceeded 2 standard errors of the mean difference. One individual was taking a beta-blocker, and his submaximal HR was likely lower during the stages of the submaximal exercise test. The second individual was an older adult ultramarathon runner, and his submaximal HR was extremely low. In both cases, the TBRS submaximal exercise test overpredicted their peak V̇o2. When using the TBRS submaximal exercise test and prediction equation, consideration must be given to the individual's medical history and physical activity level. Although the participants had a variety of chronic conditions and were taking medications for BP, cholesterol, and thyroid conditions, this submaximal exercise test has not been specifically tested in people who experience difficulty with ambulation or use an assistive device for mobility, frail elderly people, and non–community-dwelling individuals.
Because the TBRS submaximal exercise test uses the YMCA protocol, it has 3-minute stages in which outcomes such as HR, BP, and RPE can be recorded. These outcomes may be of value to the health care provider or fitness specialist to assess cardiovascular response or the individual's perceived effort (RPE) at standardized workloads (in watts) throughout the TBRS submaximal exercise test. In addition to our work, Mendelsohn and colleagues4 also used the YMCA protocol with the TBRS in 18 older adults who were living in long-term care centers. They tested whether the METs generated by the TBRS across a range of submaximal exercise intensities were similar to the METS using a portable metabolic cart. Their results suggested that data generated by the TBRS are reliable. However, they did not use the submaximal exercise test for prediction of peak V̇o2. To our knowledge, this is the first study to cross-validate the TBRS submaximal exercise test to predict peak V̇o2 in older adults. Although many participants had multiple chronic conditions, these individuals came to our laboratory interested in learning about their cardiopulmonary fitness level. In order to be included in the study, they had to reach a respiratory exchange ratio of ≥1.1 on their peak exercise test, which indicates a good effort.8 Although the difference between the actual and predicted mean peak V̇o2 values was less than 1 mL·kg−1·min−1, the TBRS submaximal exercise test did overpredict cardiopulmonary fitness. Health care and fitness professionals should consider this finding when interpreting submaximal performance using the TBRS submaximal exercise test prediction equation to estimate peak V̇o2.
Although the TBRS submaximal exercise test requires use of an HR strap and the NuStep T5XR, there are other submaximal exercise tests that require less equipment. For example, the Six-Minute Walk Test (6MWT) is commonly used for assessing physical function (distance walked). The 6MWT is easy to conduct and requires minimal equipment. Rikli and Jones15 studied 37 older adults (age range=60–87 years) to determine whether 6MWT scores would be related to V̇o2 at 85% of maximal HR. The correlation between the distance walked and peak V̇o2 was r=.78. This finding is similar to those of individuals with cardiopulmonary disorders where the distance walked is related to peak V̇o2.16,17 In 12 community-dwelling individuals with stroke, the relationship between peak V̇o2 using a cycle ergometer and V̇o2 measured during the 6MWT was r=.66 (P<.01). It is important to note that the study did not use a predictive peak V̇o2 equation but used a metabolic cart to measure V̇o2 for the duration of the 6MWT. The authors used other cardiovascular measures such as HR, rate pressure product (RPP=systolic BP × HR), and BP as indirect measures of performance during the 6MWT, but none of these measures were significantly correlated.
The Bruce protocol is a maximal-effort exercise test.9 In this seminal article in 1973, Bruce and colleagues developed several prediction equations at submaximal effort of the Bruce protocol for men and women who are healthy and men with cardiac disease. Using their population-specific predictive equations, the correlations between measured and predicted maximum V̇o2 values were .94 for men who were healthy, .93 for women who were healthy, and .87 for men with cardiac disease. These correlations are higher than our reported values. This difference may have been due to the fact that both tests were performed on the treadmill using an identical protocol for both tests, whereas we used a treadmill for obtaining peak V̇o2, and we used a seated device such as the TBRS. However, the protocol and prediction equation provide a reasonably accurate estimation (r=.87, P<.01) of peak V̇o2 for adults 60 to 80 years of age, and we believe this could be a valuable resource for clinicians, health care or fitness professionals, or researchers who want to use the TBRS.
Conclusion
The findings of the current study suggest that the TBRS submaximal exercise test and prediction equation can be used in older adults for predicting peak V̇o2. This submaximal exercise test may be useful to physical therapists and other fitness professionals who want to assess cardiopulmonary fitness but do not have access to a metabolic cart or physician supervision to conduct a maximal exercise test. The TBRS submaximal exercise test is easy to administer and can be used in a variety of settings such as fitness centers and rehabilitation settings.
Footnotes
Dr Herda and Dr Billinger provided concept/idea/research design and project management. All authors provided writing and data collection and analysis. Dr Billinger provided fund procurement and facilities/equipment. Dr Lentz provided study participants. Dr Herda provided clerical support. Dr Herda and Dr Lentz provided consultation (including review of manuscript before submission).
The study was approved by the Human Subjects Committee at the University of Kansas Medical Center.
Dr Billinger was supported, in part, by grant K01HD067318 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Ms Mattlage and Mr Sisante were supported, in part, by grant T32HD057850 from the Eunice Kennedy Shriver National Institute of Child Health and Human Development. Research in the exercise and cardiovascular health laboratory space was supported by the Georgia Holland Endowment Fund.
- Received July 17, 2013.
- Accepted January 8, 2014.
- © 2014 American Physical Therapy Association
References
Issue highlights
