Agreement of Air Bike and Treadmill Protocols to Assess Maximal Oxygen Uptake: An Exploratory Study

Abstract

Maximal oxygen consumption (V̇O_2max) is an important measure of aerobic fitness, with applications in evaluating fitness, designing training programs, and assessing overall health. While treadmill assessments are considered the gold standard, airbikes (ABs) are increasingly popular exercise machines. However, limited research exists on AB-based V̇O2max assessments, particularly regarding agreement with treadmill graded exercise tests. To address this gap, a randomized crossover study was conducted, involving 15 healthy adults (9M, 6F, 7 familiar with AB) aged 30.1 ± 8.6 years. Paired t-tests, intraclass correlation coefficients (ICC), Bland-Altman and Principal component (PC) analyses were used to assess agreement between protocols. The results demonstrated good to excellent agreement in V̇O_2max, maximum heart rate (HR), and rating of perceived exertion (ICC range: 0.89–0.92). However, significant differences were observed in several measures, including V̇O_2max and maximum HR (p < 0.01). Overall a systematic bias 3.31 mL/kg/min (treadmill > AB, 95%CI[1.67,4.94]) was observed, no proportional bias was present; however, regular AB users (systematic bias: 1.27 (95%CI[0.20,2.34]) mL/kg/min) exhibited higher agreement in V̇O_2max measures compared to non-regular users (systematic bias: 5.09 (95%CI[3.69,6.49]) mL/kg/min). There were no significant differences in cardiorespiratory coordination, between the AB and the treadmill. These findings suggest that for individuals familiar with the AB, it can be a suitable alternative for assessing V̇O_2max compared to the treadmill. Future research with larger samples should focus on developing prediction equations for field AB tests to predict V̇O_2max. Practitioners should consider using the AB to assess V̇O_2max in individuals who prefer it over running.

INTRODUCTION

A maximal assessment on a treadmill is widely considered the gold standard for assessing aerobic fitness, as determined by V̇O_2max (23). However, additional modes of assessment could be safer for maximal exercise testing as there is a risk of injury from falling off a treadmill and more viable for injured populations who cannot withstand impact (10, 28). Different exercise modes, such as swimming, the cycle ergometry, and arm-crank ergometer protocols exercise, tend to yield lower V̇O_2max values compared to the treadmill assessment in the general population (29). However, it elite swimmers and cyclists have demonstrated similar V̇O_2max values during a swimming, cycle ergometer, respectively, when compared to their treadmill V̇O_2max protocol values (29). Ultimately, each mode of V̇O_2max testing has its own advantages and disadvantages, and the choice of mode may depend on the individual’s preferences and physical capabilities. A non-treadmill running V̇O_2max protocol would be valuable as many individuals do not frequently engage in running or may have inefficient running technique (11).

An airbike (AB), also known as fan bike, is a type of stationary exercise bike that uses air resistance. ABs feature a large fan or fan blades at the front of the bike that generates resistance and are designed to spin faster as you pedal harder, creating more resistance and requiring more effort. ABs typically have dual-action handlebars that move back and forth; therefore the participant engages their upper and lower body for a full-body workout (57). In recent years, ABs have become popular in various groups, including CrossFit, the military, professional sports teams, and the general public. However, there is limited research focused on V̇O_2max assessments utilizing AB. Most existing studies have examined the effects of AB training on increases in V̇O_2max (8) and cardiorespiratory adaptations through sprint interval training on an AB (12). Kokinda and colleagues compared V̇O_2max values obtained from an Ergoline Ergoselect 200P cycle ergometer and an AirBike Renegade Pro in 5 power athletes, showing that the AB assessment was shorter in duration and yielded slightly higher median V̇O_2max values (~3 ml/kg/min) compared to the cycle ergometer (24). The small number of participants, specificity of the participants’ training, and lack of comparison to treadmill-based V̇O_2max test are key limitations of the study (24). Establishing a valid and reliable AB V̇O_2max testing protocol that adheres to standard guidelines is of value considering the popularity of the AB and the importance of aerobic fitness. Such a protocol would provide an alternative testing option for individuals who prefer not to or cannot run on a treadmill (29). Moreover, developing an AB V̇O_2max protocol would enable individuals to use familiar training equipment and tailor their training based on specific heart rate (HR) zones to enhance their ventilatory threshold, V̇O_2max, and overall aerobic conditioning.

The purpose of the present study was to develop and assess a new protocol for assessing V̇O_2max on an AB. Within this overall purpose, the first aim of this study was to investigate the agreement of an individual’s V̇O_2max during a conventional treadmill assessment and an AB assessment. Given that ABs engage both upper and lower body muscles, it was hypothesized that the AB protocol would yield V̇O_2max values comparable to the treadmill protocol (29). Secondly, based on the previously mentioned effects of training experience on V̇O_2max protocol validity (29, 30), the effect of participants’ familiarity with the AB on the agreement of V̇O_2max values with the treadmill assessment was investigated. It was hypothesized that those familiar, regularly using the AB, would display better agreement between the V̇O_2max on the AB with a treadmill.

METHODS

Participants

Fifteen participants (6 females, 9 males) completed the experimental procedures. To be eligible for the intervention, participants were required to: (1) be healthy adults 18–45 years of age; (2) not have surgery or injury in the last 3 months; (3) ability to run or bike without pain; (4) no history of cardiovascular, respiratory, or musculoskeletal diseases or use of medications that affect the responsiveness of the cardiovascular system; and (5) engage on average in 30 min of physical activity daily. All participants were informed of the benefits and risks of the study and signed the informed consent. The sample size was arrived at based on a combination of an a priori power analysis (17), prior literature (9, 24, 32) and practical limitations (27) due to resources (i.e., unfunded project, scheduling logistics, and lack of compensation for participants, etc.) available for the study. An a priori power analysis was conducted using G*Power to determine the sufficient sample size. Given the exploratory nature of the study the parameters used were an alpha of 0.05, power of 0.80, a large effect size (d = 0.8) and two tails which yielded a desired sample of 15 participants (17). The study was approved by George Mason University’s Institutional Review Board (IRB #: 2028014). This research was carried out fully in accordance to the ethical standards of the International Journal of Exercise Science (31).

Protocol

A randomized, controlled crossover design was employed for this study, where participants performed VO_2max protocols on an AB and treadmill during two testing sessions (Figure 1). Prior to the testing, participants were instructed to refrain from engaging in intense exercise for 48 hours and to abstain from eating or drinking for 2 hours, although water intake was allowed after the body composition assessment. Upon arrival at the laboratory, participants provided informed consent and completed electronic questionnaires related to their frequency of AB use, mood, personality, and lifestyle behaviors. Resting heart rate and anthropometric measures, including height, mass, and body composition, were recorded in the first data collection session. Subsequently, participants were randomly assigned to either the AB or treadmill VO_2max protocol and completed the corresponding assessment. During the second data collection session, participants performed the alternative VO_2max protocol. Standard instructions were provided to all participants, who were also familiarized with the assessments beforehand. During the familiarization, participants were provided demonstrations of the assessments, performed by the researcher, and required to perform at least 2 practice attempts to demonstrate proficiency performing the assessments. The researcher used their professional discretion if any participants needed more practice attempts. To mitigate the short-term effects of fatigue on V̇O_2max and minimize the possibility of true changes in V̇O_2max between testing sessions, all participants were required to return for the second session within a window of 48 hours to 10 days (21, 35, 38). The second V̇O_2max test was performed during the same time of day (± 2 hours) to avoid any large effects of circadian rhythm on performance (14). The testing sessions took place in the same exercise physiology laboratory, where at least 1 researcher held a certification as a National Strength and Conditioning Association Certified Strength and Conditioning Specialist (CSCS).

Figure 1.

Overview of experimental procedures.

Note: There were 6.5 ± 1.7 (M ± SD) days between day 1 and 2 for participants.

Height and mass were recorded using a stadiometer (Detecto, Webb City, MO, USA) and scale (BOD POD model 2000A; Cosmed USA, Concord, CA, USA). Percent body fat, fat mass, and fat-free mass were measured using air displacement plethysmography (BOD POD model 2000A; BOD POD, COSMED USA, Concord, CA, USA) following standardized procedures by the manufacturer. Prior to testing, the BOD POD was warmed up and calibrated according to the manufacturer’s procedures (18). Air displacement plethysmography has been shown to be a reliable and valid method of assessing body composition (15).

A novel AB V̇O_2max testing protocol was developed through pilot testing that incorporated key elements of a graded exercise test, including a warm-up period, a test duration of 8–12 minutes, and a linear increase in intensity. Two separate AB protocols were employed in this study for male and female participants, respectively, to equate the times to exhaustion. Specifically, male participants maintained a power output of 100 watts for the first 60 seconds, followed by a 25-watt increase each minute until they could no longer maintain the required wattage. Female participants, began with a power output of 80 watts and increased by 20 watts per minute until they could no longer maintain the required power output. In our pilot testing, we opted to increase the intensity of each stage by a power output (males: 25 watts, females: 20 watts) rather than RPM, as AB power output has a non-linear relationship with RPM. By increasing the intensity through power output rather than RPM the current study ensured consistent increases workload whereas small changes in RPM drastically affects the change in power output on a Rogue Echo AB (model used in our study). For treadmill testing, participants underwent the Wellness-Fitness Initiative (WFI) Treadmill ramp protocol, which has been validated in tactical athlete populations (16). The protocol started with 30 seconds at 1.5 mph followed by 3 minutes at 3 mph. Subsequently, the intensity of the protocol increased every minute by alternating between increments in speed (0.5 mph) and incline (2% increase).

During both protocols, physiological variables were continuously monitored and recorded during all stages of the test. Indirect calorimetry was conducted with a calibrated metabolic cart (TrueOne 2400, Parvo Medics, Salt Lake City, UT, USA) and metabolic data were obtained in 15 second intervals. HR was continuously recorded with a wearable chest strap monitor (H10, Polar-Electro, Kempele, Finland). Exertion was self-reported at each stage of the test via the 6 to 20 Borg rating of perceived exertion (RPE) scale (58). Prior to that start of the graded exercise test, participants were shown a visual of the 15-point RPE scale and were verbally instructed that 6 was considered extremely easy and 20 was maximal effort. In the last 30 seconds of each stage, the researcher held up the RPE visual and participants were instructed to point or verbally indicate the number that corresponded with their physical efforts. The termination criteria of VO₂ plateau, respiratory exchange ratio (RER) of 1.10 or higher, and age-predicted HR maximum were monitored for both protocols (1, 4, 29). The criteria were used to determine valid maximal effort by participants and termination of the test occurred when participants either signaled they desired to stop or physically stopped.

Statistical Analysis

A combination of histograms and Shapiro-Wilks tests for normality were used to examine the distribution of the data. Descriptive characteristics were computed for all demographics and measures from the V̇O_2max testing. Individual paired t-tests were used to determine whether significant differences in measures existed between the modes of V̇O_2max testing for all participants (n = 15). Individual paired t-tests were also used to determine whether significant differences in measures existed between the modes of V̇O_2max for a subgroup of those who reported regular use of the AB (n = 7). For the purposes of the study, regular use of the AB was defined to be using the AB at least 1 time per week for the past 6 months. To further assess agreement between the treadmill and AB, intraclass correlation coefficients (ICC)(25) and Bland-Altman plots were utilized (5). Confidence intervals for ICC estimates, bias and limits of agreements were also calculated to determine the precision of estimated limits of agreement (6). ICC were interpreted as poor (ICC < 0.5), moderate (ICC ≥ 0.5 to ICC < 0.75), good (ICC ≥ 0.75 to ICC < 0.9), and excellent (ICC ≥ 0.9). Proportional bias was assessed by linear regression between the differences in the results obtained with each VO₂ protocol with the V̇O_2max from the WFI treadmill protocol.

To assess the agreement between V̇O_2max tests on an AB with a treadmill, the network physiology framework (3, 22) may provide greater insight than more traditional measures of agreement, such as intraclass correlation coefficients (25) and Bland-Altman analyses (5). The theory of network physiology of exercise (NPE) provides a comprehensive framework to understand the non-linear dynamic interactions among different physiological systems, such as the synergism between the cardiovascular and respiratory systems during exercise stimuli (3, 22). Traditional analyses, such as correlation and Bland-Altman analyses of assessing agreement of V̇O_2max testing are limited in addressing these complex interactions as they are focused on linear relationships between measures. The NPE framework, on the other hand, utilizes non-linear modeling and time series analysis to investigate the coordination and synchronization of physiological systems (34). A method within this NPE framework, called cardiorespiratory coordination (CRC), assesses the synergism between cardiorespiratory variables using principal component analysis (PCA) (2, 19, 32). Papadakis and colleagues recently found the CRC method was able to distinguish novice and intermediate collegiate rowers completing maximal aerobic tests to exhaustion on a rowing ergometer (32). For CRC analysis, the sklearn library in Python was used to conduct PCA analyses on time series data of ventilation (VE), fraction of expired oxygen (FeO₂), fraction of expired carbon dioxide (FeCO₂), and HR (32), excluding variables such as oxygen ventilatory equivalent (VEqO₂), carbon dioxide ventilatory equivalent (VEqCO₂), oxygen (O₂) pulse, RER, and VO₂ due to their linear relationship with the selected variables. The suitability of the PCA implementation was assessed through Bartlett’s test for sphericity and the Kaiser Mayer-Olkin (KMO) test. The number of significant PCs was determined using the Kaiser-Gutmann criterion (eigenvalues > = 1.00). Due to the linear nature of the longitudinal data for HR, VE, FeO₂ and FeCO₂, missing values for data were accounted for using K-Nearest Neighbor (k = 5) method (62). There were 2 participants who had an issue with the heart rate monitor during testing and the signal was not recorded continuously; however, the values at the end of test were obtained. Further, to account for outliers, data were winsorized at the 99% CI (20). The data (HR, VE, FeO₂, FeCO₂) were then normalized using the StandardScaler function in sklearn prior to performing the PCA analysis. CRC comparisons between the treadmill and AB protocols were performed on PC1 eigenvalues using the Wilcoxon signed rank test and number of PCs using Chi-Square goodness of fit test. Effect size (Cohen’s d) was calculated when applicable to indicate standardized mean differences (26, 36). Cohen’s d values were interpreted as negligible (d < 0.2), small (d = 0.2–0.5), medium (d > 0.5–0.8) and large (d > 0.8) (13). A Benjamini-Hochberg false discovery rate of 10% was used to control for multiple analyses (37). All CRC analyses were conducted in Python 3.11.0 while all other analyses were completed using the R Environment (R Foundation for Statistical Computing, Vienna, Austria). Statistical significance was set at α < 0.05.

RESULTS

Participant (n = 15) demographics and anthropometrics are provided in Table 1. Regular AB users were found to be older (Mean = 9.5 years, Standard Deviation = 6.7, W = 8.5, p = 0.027, d = 1.27); however, there were no differences in any of the anthropometric measures between AB regular and non-regular users. All subsequent analyses met the Benjamini-Hochberg false discovery rate of 10%.

Table 1.

Participant characteristics.

	All (n = 15)			AB Regular Users (n = 7)	AB Non-Regular Users (n = 8)
	Mean (SD)	Minimum	Maximum	Mean (SD)	Mean (SD)
Age	28.9 (8.3)	21	53	34.8 (10.5)	25.3 (2.7)
Height (cm)	172.5 (7.8)	152.5	186.5	173.5 (7.4)	174.0 (5.2)
Body Mass (kg)	74.6 (15.5)	54.1	110.3	76.1 (12.0)	75.9 (17.5)
BF (%)	19.0 (6.0)	10.2	30.8	17.0 (5.0)	21.0 (7.0)
FFM (kg)	63.5 (11.2)	44.2	86.6	63.3 (12.7)	65.7 (9.0)
FM (kg)	15.9 (9.1)	8.4	40.7	12.8 (3.2)	18.6 (11.4)
Resting HR (bpm)	62.3 (11.8)	47	91	59.2 (16.0)	64.9 (9.2)

Agreement measures between the V̇O_2max protocols were evaluated using several statistical analyses, including ICC (Table 2). The ICC results demonstrated good to excellent agreement in V̇O_2max (ICC = 0.92 [0.32, 0.98], F(14,14) = 27, p < 0.001), maximum HR (ICC = 0.89 [0.49, 0.97], F(14,14) = 14, p < 0.001), and RPE (ICC = 0.91 [0.74, 0.97], F(14,14) = 11, p < 0.001). However, paired t-tests indicated significant differences between the V̇O_2max protocols in several measures. Specifically, there were significant differences observed in V̇O_2max (t(14) = 4.344, p < 0.001), maximum HR (t(14) = 3.137, p = 0.007), HR at ventilatory threshold (t(14) = 3.543, p = 0.003), and test duration (t(14) = 5.572, p < 0.001). Furthermore, Bland-Altman analyses revealed a systematic bias of 3.31 mL/kg/min (treadmill > AB, 95%CI[1.67, 4.94]), with a lower limit of agreement of −2.59 (95%CI[−5.42, 0.24]) and an upper limit of agreement of 9.20 (95%CI[6.38, 12.03]). Notably, no proportional bias was observed between the two protocols (Figure 2).

Table 2.

Comparison of VO_2max measures between treadmill and airbike protocols.

			Paired t-test		Intraclass Correlation Coefficient
	Treadmill M (SD)	Airbike M (SD)	P-value	ES	ICC [95% CI]	p-value
VO2max (mL/kg/min)	45.1 (7.5)	41.8 (7.8)	< 0.001	0.42	0.92 [0.32,0.98]	< 0.001
RER	1.16 (0.06)	1.13 (0.04)	0.113	0.52	0.40 [−0.54,0.79]	0.15
Max HR (bpm)	185.1 (10.9)	182.3 (9.5)	0.007	0.42	0.89 [0.49,0.97]	< 0.001
Final RPE	19.0 (1.3)	19.1 (1.3)	0.498	0.10	0.91 [0.74, 0.97]	< 0.001
VO2max at Ventilatory Threshold (%)	68.0 (9.0)	59.0 (17.0)	0.065	0.64	0.34 [−0.58, 0.76]	0.19
HR at Ventilatory Threshold (bpm)	151.9 (14.9)	130.9 (28.0)	0.003	1.00	0.43 [−0.28, 0.78]	0.064
Duration (s)	731.7 (100.9)	567.7 (135.3)	< 0.001	1.38	0.44 [−0.26, −0.80]	0.016
CRC Eigenvalue	2.51 (0.27)	2.82 (0.53)	0.132	0.73	-	-

Notes:

1. The p-value and effect sizes are results from paired t-tests.
2. Eigenvalues were compared with a Wilcoxon signed rank test.
3. The ICC(2,1), intraclass correlation coefficients, are presented in the final column with 95% confident intervals. ICC were not computed for the eigenvalues.
4. The treadmill protocol included a 3 minute (180 second) warm-up period).
5. Abbreviations: M, mean; SD, standard deviation; ES, effect size; CI, confidence interval; CRC, Cardiorespiratory Coordination.

Figure 2.

Bland-Altman plot of agreement for treadmill versus airbike VO_2max protocols for all participants.

There was no difference in V̇O_2max on the AB between participants who reported regular use of the AB compared to those who did not use the AB regularly (W = 13.5, p = 0.105, d = 0.80). The only V̇O_2max test parameter found to differ between the regular and non-regular AB users was test duration (regular user > non-regular user duration; W = 9, p = 0.029, d = 1.26). However, subsequent analyses (Table 3) revealed a higher level of agreement in V̇O_2max measures among participants who regularly use the AB (Figure 3) compared to those who do not (Figure 4). Specifically, a systematic bias of 1.27 (95%CI[0.20, 2.34]) mL/kg/min was observed among regular AB users, with a lower and upper limit of agreement of −2.60 (95%CI[−4.46, −0.74]) and 5.15 (95%CI[3.29, 7.00]) mL/kg/min, respectively. Whereas, participants who did not regularly use the AB exhibited a systematic bias of 5.09 (95%CI[3.69, 6.49]) mL/kg/min, with a lower limit of agreement of 0.03 (95%CI[−2.40, 2.45]) mL/kg/min and an upper limit of agreement of 10.15 (95%CI[7.72, 12.58]) mL/kg/min. No proportional bias was found for either group based on AB user experience. Notably, no significant differences were found in V̇O_2max, RER, or maximum HR obtained from the maximal tests performed on the treadmill and AB for participants who regularly use the AB (Table 3).

Table 3.

Comparison of VO_2max measures between treadmill and airbike protocols for regular and non-regular airbike users.

	Regular Airbike Users (n = 7)				Non-regular Airbike Users (n = 8)
	Treadmill Median (Q1–Q3)	Airbike Median (Q1–Q3)	P-value	ES	Treadmill Median (Q1–Q3)	Airbike Median (Q1–Q3)	P-value	ES
VO2max (mL/kg/min)	47.8 (41.0–51.0)	46.2 (39.3–48.3)	0.219	0.20	43.75 (40.30–50.05)	36.95 (35.83–45.4)	0.008	0.58
RER	1.15 (1.11–1.16)	1.13 (1.11–1.16)	0.932	0.07	1.19 (1.14–1.22)	1.13 (1.12–1.15)	0.042	0.92
Maximum HR (bpm)	186.0 (173.0–192.0)	187. (172.5–192.5)	0.400	0.08	189.0 (186.0–191.2)	180.5 (179.0–185.0)	0.016	1.26
Final RPE	20.0 (19.0–20.0)	20.0 (19.0–20.0)	1.000	0.00	19.0 (17.0–20.0)	19.5 (17.8–20.0)	0.484	0.15
VO2max at Ventilatory Threshold (%)	72.0 (71.5–76.5)	64.0 (47.0–74.0)	0.078	0.96	62.5 (59.5–66.3)	51.5 (49.5–57.3)	0.383	0.40
HR at Ventilatory Threshold (bpm)	158.0 (151.5–170.0)	136.0 (129.5–162.5)	0.078	0.86	148.5 (141.2–155.8)	112.0 (101.8–129.2)	0.039	1.26
Duration (s)	788.0 (658.5–824.0)	624.0 (594.0–652.0)	0.078	1.00	762.0 (678.8–788.5)	491.0 (441.5–579.0)	0.008	1.87
CRC Eigenvalue	2.59 (2.28–2.68)	2.59 (2.53–3.13)	0.208	0.94	2.47 (2.38–2.74)	2.95 (2.31–3.34)	0.313	0.55

Abbreviations: IQR, interquartile range; ES, effect size; HR, heart rate; RPE, rating of perceived effort; VT, ventilatory threshold; CRC, Cardiorespiratory Coordination. Note: Due to small sample sizes the non-parametric Wilcoxon signed rank test (paired samples) was used to assess differences between the treadmill and airbike measures.

Figure 3.

Bland-Altman plot of agreement for treadmill versus airbike VO_2max protocols for regular airbike users.

Figure 4.

Bland-Altman plot of agreement for treadmill versus airbike VO_2max protocols for non regular airbike users.

There were no significant differences (V = 28, p = 0.132) in the eigenvalues of PC1 (Table 2), which accounted for the highest proportion of the data variance, between the AB (Mean = 2.82; SD = 0.53) and the treadmill (Mean = 2.52; SD = 0.27; Supplementary Table 1). However, there were significant differences in the projections of HR (treadmill > AB, V = 91, p = 0.017, d = 1.20) and fraction of expired oxygen (treadmill < AB, V = 6, p = 0.004, d = 1.27) in PC1. Notably, a greater number of participants (n = 15) exhibited a second principal component for the treadmill (86.7%, n = 13) compared to the AB (46.7%, n = 7); however, a chi-square goodness of fit indicated no significant difference in the number of PCs (X²(1, N = 15) = 3.75, p = 0.053). Further analyses were performed to examine possible disparities in the principal component analysis (PCA) outcomes between and within individuals who regularly use the AB and those who do not. However, none of the PCA metrics, including eigenvalues and PC1 projections, exhibited statistically significant differences between experienced AB users and inexperienced individuals. Thus, the CRC results provided additional insight than the traditional ICC and Bland-Altman analyses. Suggesting that while the overall responses to each protocol were similar there were some subtle differences in the dynamical coordination of the cardiovascular and respiratory systems for some, but not all, participants.

DISCUSSION

The primary aim of this study was to examine the agreement between V̇O_2max values obtained from a traditional treadmill assessment and an AB assessment, while also investigating the potential impact of familiarity with the AB on this agreement. Our hypotheses were partially supported by the findings. Significant differences were observed in V̇O_2max, maximum HR, HR at ventilatory threshold, and duration between the two assessment methods. However, there were no differences in CRC metrics, indicating a comparable response of the cardiorespiratory system to both protocols. Notably, among individuals who regularly use the AB, a stronger agreement was observed between the treadmill and AB results, supporting previous literature emphasizing the importance of considering an individual’s training history when selecting the mode of V̇O_2max testing (30). Thus, the findings suggest that while there is insufficient evidence to support the use of the AB for assessing V̇O_2max in all individuals, it appears to be a suitable alternative to treadmill testing for those who are familiar with the AB.

As previously mentioned, prior research (24) compared the AB to cycle ergometer protocol and found a median difference of about 3 mL/kg/min in V̇O_2max values, with the AB testing producing greater values than the cycle ergometer. In addition to differences in sample size and inclusion criteria, another notable distinction between our study was the duration of the AB protocols (24). This prior study (24) reported an average AB protocol duration of 252 seconds (~4.2 minutes), while our study had an average AB protocol duration of 567.7 seconds (~9.5 minutes). This is likely attributed to the methodology of the previous study (24), which used a protocol starting at a baseline pedaling speed of 55 rpm and increasing by 3 rpm every 20 seconds. Considering the recommended duration for a V̇O_2max test of 8 to 12 minutes (29), the protocol used in our study appears to be more appropriate for a broader population. Another study (9) compared the agreement between V̇O_2max tests on an AB and a cycle ergometer and found that the AB testing produced a greater V̇O_2max than on a cycle ergometer by about 9% (AB V̇O_2max: 53.1 ± 8.7 mL/kg/min; Cycle ergometer V̇O_2max: 47.4 ± 9.2 mL/kg/min) in 18 physically active young males men. Notably, Canário-Lemos et al. (9) had participants start at an RPM of 30 (equivalent to 27 watts) and increase by 5 rpm every minute on the AB, with a mean AB testing duration of 598 seconds (~10 minutes), closer to the duration in our study (567.7 seconds). This approach ensured a more graded test and prevented premature termination of the AB testing. Collectively, these studies offer contrasting viewpoints on the validity of the AB protocol for V̇O_2max assessment (9,24). The divergent populations and protocol durations indicate the need for additional research to establish a comprehensive understanding of the AB’s suitability as a method for measuring V̇O_2max.

Interestingly, while some of the V̇O_2max test metrics showed similarities between the protocols (i.e., RER, final RPE, %V̇O_2max at ventilatory threshold, CRC), other metrics exhibited differences (i.e., maximum HR, HR at ventilatory threshold; Table 2). When considering all participants, the treadmill protocol elicited higher exercise intensity in terms of V̇O_2max, maximum HR, and HR at ventilatory threshold. This discrepancy could be attributed to the greater intensity of the treadmill protocol, resulting in statistically higher values for these physiological measure or the duration of the treadmill test being longer. However, it is important to note that the difference in test duration between the treadmill and AB protocols in this study may be misleading, as the WFI protocol included a 210-second walking warm-up (8). When considering the actual exercise duration, the treadmill duration without the warm-up was 521.7 seconds, which is closer to the average duration of the AB protocol (567.7 seconds), but still significantly shorter. Participants subjectively reported that the initial minutes of the AB protocol were primarily a warm-up based on their RPE. Another plausible explanation for the higher HR-related measures during the treadmill protocol is the documented difference in HR response between running and cycling at similar intensities (30). During maximal exercise testing, the HR response has typically been found to be approximately 5% lower during an incremental cycle test compared to a treadmill test (30, 33). Training specificity is a factor known to influence HR response during maximal exercise testing (30), which leads to what we believe explains the differences in results for AB and non-AB users of this study.

Although there was a significant difference in V̇O_2max between the treadmill and AB protocols for all participants, these differences tended to be primarily driven by individuals who were not regular AB users. It is widely recognized that untrained individuals or recreational athletes typically achieve higher V̇O_2max values on maximal treadmill tests compared to other modes, such as cycling (29). Conversely, highly trained individuals who specialize in a specific mode of exercise generally attain higher V̇O_2max values in their trained mode (30). Bouckaert et al. conducted a study on highly trained runners and cyclists, finding a 14% higher V̇O_2max on the treadmill for runners, while cyclists had an 11% higher V̇O_2max on the bicycle ergometer (7). Triathletes, who engage in multiple modes of exercise, offer a unique opportunity to investigate V̇O_2max. Consistent research findings indicate that regular training in multiple modes results in similar V̇O_2max values, regardless of the testing mode employed (30). Our findings further emphasize the importance of considering training history and associated adaptations when selecting testing protocols and interpreting V̇O_2max results.

The present study has several limitations. First, the sample of 15 healthy adults, including both sexes and a wide age range of 21 to 53, may restrict the generalizability of the findings to specific populations of interest. To enhance the applicability of the results, future research should use larger samples and include participants from specific populations relevant to the study’s objectives. Furthermore, developing a prediction equation for estimating V̇O_2max from a field test on the AB would be valuable for practical applications. A larger sample size would allow for the development of a robust prediction equation incorporating multiple predictors such as age, sex, body mass, and HR. Lastly, it is important to note that our assessment of regular AB users was based on self-reporting, and the categorization criterion of at least once per week may be considered as a crude measure. Future studies would benefit from more precise quantification of AB usage to accurately categorize participants as regular or non-regular users.

The AB protocol showed promise as a valid assessment of V̇O_2max when compared to the treadmill protocol in individuals who regularly use the AB. These results suggest that the AB protocol can be effectively utilized as a measure of V̇O_2max in specific populations, such as athletes or individuals who frequently engage in AB exercises as part of their training regimen. Furthermore, the AB protocol showed similar values of the RER, final RPE, and percent V̇O_2max at the ventilatory threshold, indicating its ability to provide valuable information on metabolic responses and exertion levels during exercise. However, practitioners should exercise caution when using the AB protocol interchangeably with treadmill-based V̇O_2max testing, as significant differences were observed between the two protocols.