An a priori power analysis was conducted using G*Power (version 3.1.9.3) (23) to determine the minimum number of participants required to detect a potential difference in VO_2max between the treadmill and RA test. Assuming a paired-samples t-test, a two-tailed α level of 0.05, statistical power (1 – β) of 0.90, and a small-to-moderate effect size (d ≤ 0.5), the analysis indicated a minimum sample size of 44 participants (critical t = 2.02, df = 43). With 53 participants included in the final analysis, the study exceeded this requirement, achieving an actual power of 0.90 and a minimum detectable effect size of d = 0.41, thus ensuring sufficient power to detect meaningful differences in VO_2max between protocols.

Sixty participants from the sport science student community of the university and local sport clubs were initially recruited for the study. Seven were excluded from the analysis due to missing data, failure to complete both tests, and, due to not reaching VO_2max during the treadmill test. The final sample contained 53 participants (27 females, 26 males) The participant characteristics are summarized in Table 1.

Participants were trained team ball sport athletes (football, handball, volleyball, hockey, basketball) corresponding to tier 2 and tier 3 according to the classification of McKay (24). Participants had on average 12.1 (±6) years of training experience, performed 4.1 ± 1.6 training sessions per week with a weekly training load of 6.8 (±3.1) hours. Participants with muscular injuries, cardio-vascular diseases or any other limitation on the test day (e.g., sickness) were excluded. Participants were informed about the experimental protocol, and written consent was obtained prior to testing. The study was approved by the Luxembourgish national research ethics committee (Nr. 202207/01 v2.0) and conducted in accordance with the Declaration of Helsinki.

Participants visited the lab on two occasions with at least 48 h between. To avoid influence of circadian rhythm tests were performed at the same time of the day ± 3 h. Participants were instructed to abstain from alcohol and caffeinated drinks at least 24 h prior to testing and maintain their habitual diet. Moreover, there should be no intense training on the day before the lab visit.

On day 1, participants completed an exhaustive treadmill VO_2max test followed by a familiarization trial with the RA Test on the SKILLCOURT to minimize potential learning effects. On day 2, they performed the RA Test. During all tests, gas exchange was continuously measured breath-by-breath using a validated mobile MetaMax 3B analyzer (CORTEX Biophysik GmbH, Leipzig, Germany) (25). The gas analysis system was calibrated according to the manufacturer guidelines using reference gas and ambient air calibration as well as flow sensor volume calibration with a 3 L calibration syringe. Heart rate (HR) was continuously measured using a H10 sensor (Polar Elektro, Kempele, Finland). Rate of perceived exertion (RPE; 6–20) was obtained according to Borg's scale. Blood lactate samples were taken at the earlobe and analyzed using a Biosen C-Line lactate analyzer (EKF-diagnostic GmbH, Barleben, Germany).

The ramp-like incremental test was performed on a treadmill (h/p/cosmos®, Pulsar®, Nussdorf, Germany). To account for differences in performance, participants started either at 6 km × h⁻¹ or 8 km × h⁻¹ with a 1% incline (26). The decision was taken based on training experience, number of weekly training sessions, training load, previous performance tests (if available) and personal rating of performance status. The protocol for the treadmill test is illustrated in Figure 1A. Participants warmed up for 3 min at the starting velocity. This was followed by a speed increase of 0.5 km × h⁻¹ every 30 s up to a velocity of 16 km × h⁻¹. Afterwards, inclination was increased by 1% per minute. Participants were verbally encouraged throughout the test. RPE and lactate were determined prior to and immediately after the test. VO₂ and HR were continuously recorded. The average test duration was 10.7 (±2.2) min. A 30s moving average was applied to the raw data and VO_2max was defined as the highest VO₂ value. VO_2max was considered as valid if the VO₂ increase during the last minute did not exceed 150 mL indicating a levelling-off (27). Alternatively, two of the four criteria must be met. 1) RER≥1.1, 2) blood lactate concentration ≥ 8mmol × l⁻¹, 3) HR ≥ 95% of maximum HR (220-age) or 4) RPE ≥ 18 (19, 28). Participants, neither reaching the primary (levelling-off) nor secondary criteria, were excluded from the analysis.

On test day 2, participants performed the RA Test on the SKILLCOURT (Skillcourt GmbH, Schweinfurt, Germany). The test contains four all-out 150 m reactive agility runs with 30 s of rest in between runs that were performed on a 4 × 4 m court. The test protocol for the RA Test is illustrated in Figure 1B. Participants started with a warm-up comprising 3 min of moderate, self-paced treadmill running followed by 5 min of stretching. Afterwards, two RA runs of 50 m were performed on the SKILLCOURT. Participants were asked to perform the runs at 60% and 80% of their individual maximum performance. The intensities were chosen to increase the physiological demands following the general warm-up and prepare participants for the maximum intensity all-out RA runs. After 2.5 min recovery, the first run started. Participants had to run to one out of eight target fields as indicated on the device screen. Once a field was reached the next showed up. The sequence was randomized, and distance was automatically calculated by the system using a LiDAR (Light Detection and Ranging). Participants were instructed to perform each run as fast as possible and received verbal encouragement throughout the test. To ensure consistency in effort and time, in the event of a RA error run, the test proceeded, and participants continued to the next target field. As for the treadmill test, RPE and lactate were determined prior to and immediately after the test. VO₂ and HR were continuously recorded during both tests. The test time for the RA Test was 8.2 (±0.5) min (6.2 min running time + 4 × 30 s break). As for the treadmill test, a 30s moving average was applied to the raw data and the highest VO₂ was considered as VO_2max.

In addition to its direct measurement, VO_2max was estimated through an equation (Equation 1) established from a previous study, which investigated the relationship between total running time in the RA Test and treadmill-based VO_2max (Karsten et al. 2025—under review).estimatedVO2max=−0.1681×RARunTotalTime(s)+115.65All metrics for the treadmill and RA Test are presented in Table 2.

Data was analyzed in JASP (version 0.19.0.3) and SPSS (version 29.0.2.0). Shapiro–Wilk tests were used to test for normal distribution and non-parametric tests were used in case of normal distribution violation. Control analyses tested for differences in resting state HR, LA, RPE and test duration.

Agreement between treadmill VO_2max and RA Test VO_2max was assessed using Limits of Agreement (LoA) according to Bland and Altman (29). Lin's concordance correlation coefficient (CCC) for paired measurements (30) and Pearson correlation coefficient were used for correlation analysis between VO_2max measured in the RA Test and treadmill VO_2max. A dependent t-test was applied to test for significant differences in measured VO_2max between the treadmill and RA Test. The intraclass correlation coefficient (ICC) was calculated based on a two-way mixed effects model with single measurement and absolute agreement (31) according to the classification of McGraw and Wong (32) to determine the degree of similarity in VO_2max between the treadmill and RA Test. The same set of analyses was performed to compare the treadmill VO_2max to the predicted VO_2max values based on the overall running time in the RA Test. To account for potential sex differences, all analyses were also carried out separately for male and female participants.

Effect sizes were considered small (d = 0.2, r = 0.1), medium (d = 0.5, r = 0.3) or large (d = 0.8, r = 0.5). CCC was considered nearly perfect [precision(p_c) > 0.99], substantial (p_c > 0.95), moderate (p_c > 0.9) or poor (p_c ≤ 0.9). The significance threshold was set to p < 0.05.

No significant difference was found between VO_2max values obtained from the RA Test and the treadmill protocol (t = 0.606, p = 0.547, d = 0.083). Bland–Altman analysis showed a mean bias of 0.25 mL·kg⁻¹·min⁻¹, with 95% LoA of +6.02 (∼11%) and −5.53 mL·kg⁻¹·min⁻¹ (∼10%), indicating a negligible underestimation (<0.5%) of VO_2max in the RA Test. The Lin's concordance correlation coefficient (ρc = 0.94) demonstrated moderate agreement between methods. A very strong Pearson correlation was observed (r = 0.94, p < 0.001), and the intraclass correlation coefficient (ICC = 0.943; 95% CI 0.907–0.966) confirmed excellent agreement between treadmill and RA Test VO_2max measurements. The findings from the overall (whole-group) analysis are depicted in Figure 2.

LoA analysis between estimated and measured VO_2max on the treadmill revealed a difference of 1.33 mL × kg⁻¹ ×  min⁻¹ (3.9%) with upper and lower 95% LoA of 12.8 (24%) and −10.1 (19%) mL × kg⁻¹ ×  min⁻¹, respectively. There was no significant difference between VO2_max measured on the treadmill and predicted form the running time in the RA Test (t = 1.610, p = 0.114, d = 0.228). Strong correlations were observed between the estimated VO_2max in the RA Test and the measured treadmill VO_2max (r = 0.74, p < 0.001). However, with a coefficient of concordance of p_c = 0.736, the agreement was rather poor. Also, the ICC of 0.674 indicated only moderate similarity between measured treadmill VO_2max and estimated values from the RA Test. Results are presented in Figure 3.

Results for the subgroups of male and female athletes are summarized in Table 3. The findings from the overall analyses were consistent across subgroups, particularly for the directly measured VO_2max in the RA test. While for the predicted VO_2max, there was a significant difference between the treadmill and RA test only for the male group (t = 2.457, p = 0.022, d = 0.491) and ICC and CCC where lower in absolute numbers in the male when compared to the female participants, effect sizes for the t-test as well as ICC and CCC values were not significantly different (p ≥ 0.171).

With a mean difference of 0.25 mL × kg⁻¹ ×  min⁻¹, corresponding to about 0.5%, there was a strong agreement in the measured VO_2max between the two tests. The agreement achieved in the RA Test was substantially stronger than those reported by Martinez-Luganas and Hartmann (18) who found the Yo-Yo IR1 underestimated treadmill VO_2max by 9.4% (95% LoA: −20% to +1.4%) in female football players. Similarly Castagna et al. (17) observed a 2.67 mL·kg⁻¹·min⁻¹ (∼5%) underestimation in male youth players, with LoA of −14 to +8.6 mL·kg⁻¹·min⁻¹ (−26% to +16%), roughly twice as wide as in the RA Test. Interestingly, the underestimation of VO_2max was substantially smaller (1.7 mL × kg⁻¹ ×  min⁻¹, 3.1%) in the YoYo-IR2 test as reported by Born et al. (19). The same applied to the incremental test on the SpeedCourt incorporating a RA component (1.7 mL × kg⁻¹ ×  min⁻¹, 3.1%). LoA were substantially larger when compared to the RA Test with about −9–9 mL × kg⁻¹ × min⁻¹ (−16% to +16%) for the YoYo-IR2 and −6 to +10 mL × kg⁻¹ ×  min⁻¹ (−10.8% to +18%) for the incremental RA test on the SpeedCourt.

Although the LoA in the present study (+11% and −10%) may still be considerable wide, this degree of variability falls within the known biological and technical error of VO_2max testing. Katch et al. (33) reported a total error of 5.6% across repeated maximal tests. An additional 2% error is attributable to gas-analyser variability (34). Hence, the combined error margin (∼7%) supports the conclusion that the RA Test demonstrates good validity for eliciting VO_2max.

The correlation between VO_2max measured on the treadmill and in the RA Test (r = 0.94) was very high and significantly stronger when compared to the correlation reported between treadmill VO_2max and the YoYo-IR1 test by Martinez-Luganas and Hartmann (18) (r = 0.94 vs. r = 0.83, p = 0.039). The study by Castagna et al. (17) observed an even lower correlation between treadmill and YoYo-IR1 VO_2peak of r = 0.65. Interestingly, while the agility-like test on the SpeedCourt was closest in reaching the treadmill VO_2max, the correlation between SpeedCourt and treadmill VO_2max was comparatively low (r = 0.59) (19). When considering the strong correlation between the treadmill and RA Test together with the very high concordance correlation coefficient (p_c = 0.94) and ICC (0.94), these findings confirm excellent agreement between VO_2max values measured during the treadmill and RA Test.

In contrast to other field-based protocols, the motor-cognitive RA Test reliably elicits VO_2max while replicating the reactive multidirectional, and cognitive demands of team ball sports. Accordingly, the RA Test provides a valid, sport-specific, and practically applicable option for assessing aerobic capacity in team ball sport athletes.

The predicted VO_2max based on the overall running time in the RA Test also did not indicate a significant difference from the directly measured VO_2max on the treadmill. The estimation in the RA Test demonstrated greater precision than previously reported for the YoYo test, where the VO_2max predicted using the Bangsbo et al. (15) equation underestimated the treadmill-measured VO_2max by 17.8% in female football players (18). Comparable values were reported by Michailidis et al. (35) for male football players where VO_2max was underestimated by 14% in the YoYo-IR1. The largest difference was observed by Kramer et al. (36) with an underestimation of 30% when comparing estimated VO_2max to measured VO_2max in a YoYo-IR1 test. Although the prediction of VO_2max from the RA Test was on average accurate, the LoA were substantially wider than those observed for directly measured VO_2max and comparable to previous findings for the YoYo-test (18). Considering the wider LoA between 24% and −19% together with the concordance correlation coefficient of p_c = 0.736 and the ICC of 0.68, the estimated VO_2max from the RA Test should therefore be interpreted as an approximate indicator rather than a precise prediction of VO_2max. This interpretation is further supported by the correlation analysis which revealed an association of 0.737, similar to the range typically reported for the YoYo test [r = 0.43 and r = 0.87 (16)]. Given the RA Test's reactive and multidirectional characteristics, it is plausible that CoD ability and anaerobic energy contributions influence running performance and, consequently, reduced the strength of the correlation. In fact, CoD ability does depend on strength, power and technique which may substantially vary between participants independent of aerobic capacity (37). Moreover, there was only a moderate correlation (r = −0.346) between VO_2max and overall running time in an 8 × 40 m sprint protocol (38). Therefore, although the running distance was longer in this study (4 × 150 m), a substantial contribution of the anaerobic energy metabolism to the overall running time in the RA Test can be assumed.

The motor-cognitive RA Test was specifically designed for team ball sport athletes to provide a higher stimulus-correspondence (through external visual cue) and task correspondence (through all-out multidirectional CoD) compared to existing field-based endurance tests. VO_2max values closely align with those obtained from treadmill testing and they provide a higher precision and accuracy in obtaining VO_2max when compared to established field-based endurance tests. The RA test can therefore serve as an option and alternative for athletes in team ball sports to determine aerobic capacity in a more sport-specific setting. As such, the RA test may qualify as a performance test for evaluating training programmes or athlete selection.

With a test duration of 8.2 ± 0.6 min, the RA Test is more time efficient than the treadmill testing (10.7 ± 2.2 min) and comparable in length to the YoYo test (18). Importantly, 90% of the participants reached their VO_2max within the first or second 150-m run interval, suggesting the potential to shorten the protocol to two intervals of 2 × 150-m in future applications. However, as the RA Test is an all-out test, a 5 min warm-up should be considered. Still, the total testing time was shorter when compared to the incremental agility-like test reported by Born et al. (19) (∼18–19 min). Finally, due to distance measurement by the LiDAR and automatic data processing, the test provides a high objectivity, and it is less staff-intensive when compared to e.g., a YoYo-test, although only one athlete can be tested at a time. The estimation of VO_2max values based on overall running time may be useful for coaches and athletes when gas analysis is not available but should be interpreted with caution. In this context the same limitations apply to the RA Test as to commonly used field-based VO_2max tests.

While the study confirms that the RA Test can obtain a VO_2max response, reliability has not been addressed and needs to be confirmed in future research. To support generalizability across performance levels and disciplines, additional studies with different athlete populations are warranted. Further, oxygen kinetics were not considered and should be analyzed to provide an explanation for the higher validity when compared to many existing field-based tests. Future research should also evaluate whether performance in the RA Test is related to physiological on-court performance (e.g., running distance, number of sprints, etc.) as previously indicated for the YoYo-test (15, 39, 40). Finally, although this study employed SKILLCOURT technology to measure and estimate VO_2max, similar RA test protocols could potentially be implemented using other technologies (e.g., reaction lights), which warrants evaluation in future research.