Twenty-seven high school male soccer players who were registered in the Korean Football Association participated in this study. A researcher who received prior approval from the head coach visited the training ground, explained the purpose of the reseach to the participants as well as their parents, and obtained their voluntary informed consent to participate in the study. Characteristics of the participants are shown in Table 1.

All data collection was performed at a laboratory setting or on a training ground with an artificial grass surface. Among the twenty-seven players, those without injuries performed two graded exercise tests (GXT) on a treadmill (20 of 27 players participated) before and after preseason training and the 21-m SRT on four separate occasions at the same time of day during preseason training. To determine the reliability of the 21-m SRT, it was conducted twice, a test and a retest, at a one-week interval (weeks 1 and 2) immediately after the start of preseason training (27 players participated). To evaluate its validity and practicality, it was also conducted three times at four-week intervals (weeks 2, 6, and 10) during the last eight weeks of preseason training (16 of 27 players participated). The maximal oxygen consumption of each participant was measured in the laboratory before and after preseason training to examine the validity of the 21-SRT as a tool for soccer players’ fitness development. Each procedure was configured as shown in Figure 1. Participants were instructed not to consume food or caffeine at least 4 hours prior to each test.

The reliability, validity, and practicality of the SRT with an irregular recovery period, which was developed by Lee and Yun (2005), for evaluating the aerobic capacity of soccer players in a field setting was examined.

Before the SRT, participants performed 10 minutes of light-intensity warming up with stretching. The SRT was performed on artificial turf. The SRT requires participants to run twenty-one meters back and forth across a marked track keeping time with prerecorded beep signals. If a participant reaches the relevant marker in time, 15 seconds of active recovery time, during which the participant walks within an eight-meter area, is given, then the next level commences: participants must run faster in order to keep time with faster beeps (Figure 2). If a participant fails to reach the relevant marker in time, they are warned. A second warning terminates the test for that participant. The number of shuttle runs completed successfully and the total distance traversed are recorded to represent the performance of that player. The heart rate of participants during the SRT were obtained using the AIRBEAT system [8].

Participants were taken to an air-conditioned laboratory which had a relative humidity of 50-60% at a temperature of 20-22℃. The GXT was performed on a treadmill with a modified Bruce protocol, which is highly correlated to directed measured maximal oxygen consumption [9]. Individual maximal oxygen consumption was obtained using open-circuit spirometer techniques (Quark metabolic cart, COSMED, Italy), and heart rate was measured with a Polar heartrate monitor. During the GXT, gas exchange data were collected continuously using an automated breath-by-breath method and were manipulated to calculate V_E, VO₂ consumption, and VCO₂ production; Borg’s rating of perceived exertion (RPE) was measured at the end of every minute [10]. All participants were encouraged to exercise to volitional exhaustion during the GXT. VO_2max was reached when at least two of the following criteria were satisfied: a leveling off of the VO₂, a RPE > 17 using Borg’s scale, voluntary exhaustion, and achievement of 90% the age-predicted maximal heart 4 rate [11,12].

To test its reliability, participants performed the 21-m SRT on two separate occasions, exactly 1 week apart, and at the same time of day. Pearson’s correlation coefficients and paired sample t-Figure 1. Experimental procedures. t-tests were used to examine the relationship and systemic bias between the test and retest. The criterion-related validity of 21-m SRT performance was investigated using Pearson’s correlation coefficients between directly measured VO_2max and 21-m SRT performance, which is the total distance covered during the test. To evaluate the practicality of the 21-m SRT for soccer players during preseason training, paired sample t-tests were performed for the level of VO_2max before and after 8 weeks of preseason training. Moreover, 21-m SRT performance, which is represented by heartrate immediately after the 67^th repetition (SRT_HR67) and by the total distance covered (SRT_Distance), during the preseason training was measured on three separate occasions with two 4-week intervals. Statistical comparisons were performed using a repeated one-way ANOVA design with a within-subject factor of time [baseline (2^nd week), 6^th week, and 10^th week] for 21-m SRT performance during preseason training. When a significant MF value was obtained, a posthoc test using the Bonferroni method was applied. Statistical analyses were performed using GraphPad Prism 9.5.1 packages.

The test and retest of the 21-m SRT were performed by 27 participants exactly 1 week apart at the beginning of preseason training. The mean of SRT_Distance for the test was 2770 ± 451.2 m (95% confidence interval (CI) 2591 - 2948 m) and that for the retest was 2916 ± 482.2 m (95% CI 2725 - 3107 m). A strong correlation existed between the test and retest for SRT_Distance (Figure 3A). The paired t-test conducted to test the hypothesis for the equality of mean between the SRT test and retest revealed a statistically significant bias (p = 0.0057). The absolute mean difference ± 95% Limit of Agreement was 146.2 ± 494.1 m. The Bland-Altman plot for mean test and retest is depicted in Figure 3B.

Twenty players performed a GXT on a treadmill to examine the criterion-related validity of the 21-m SRT using Pearson’s correlation coefficients between directly measured VO_2max in the laboratory and SRT_Distance performed in the second week. The mean VO_2max of the twenty players was 58.6 ± 4.6 ml・min・kg^-1 (95% CI, 56.4 - 60.7 ml・min・ kg^-1). There was a significant correlation between directly measured VO_2max and SRT_Distance (r = 0.465, p = 0.039). The equation for estimated VO_2max from the 21-m SRT is depicted in Figure 4.

Twenty players performed a GXT on a treadmill before and after 10 weeks of preseason training to examine the effect of preseason training on maximal oxygen consumption. The VO_2max of the twenty players [61.6 ± 5.3 ml・min・ kg^-1 (95% CI, 59.2 - 64.1 ml・min・kg^-1)] was significantly higher after the preseason training (t(19) = 2.36, 95% CI of difference 0.3403 to 5.780, p < 0.05). To determine the manner of the changes in players’ aerobic capacity after preseason training, whether 21-m SRT performance (SRT_Distance and SRT_HR67) differed between training periods, which were separated by 4 weeks, was examined. Among the twenty-seven participants, only sixteen successfully completed all three 21-m SRTs during preseason training without injury. There were statistically significant differences in 21-m SRT performance (F (2, 15) = 13.59, p = 0.0002, Figure 5A for SRT_Distance; F (2, 15) = 10.83, p = 0.0003, Figure 5B for SRT_HR67). A post hoc test using Bonferroni correction demonstrated that SRT_Distance at the 10^th week (3504 ± 458 m) was significantly higher than that of the 2^nd week (mean difference = 485.6 m, 95% CI of difference 222.6 m to 748.7 m, p = 0.0005) and than that of the 6^th week (mean difference = 485.6 m, 95% CI of difference 222.6 m to 748.7 m, p = 0.0005). Moreover, SRT_HR67th had significantly decreased at the 6^th week (mean difference -5.0 bpm, 95% CI -9.1 bpm to -0.93 bpm, p = 0.0143) and at the 10^th week (mean difference -6.7 bpm, 95% CI -10.8 bpm to -2.6 bpm, p = 0.0014) compared to that of the 2^nd week (177.9 ± 5.48 bpm). Moreover, there was a significant negative correlation between SRT_HR67 and SRT_Distance during the preseason training period (r = -0.311, p < 0.05, Figure 5C).