Knowledge of the inherent variation in human exercise performance is critical to the detection and valid interpretation of effects on performance associated with experimental manipulations. Information concerning the reliability of anaerobically biased, high-power exercise tests is especially limited. For example, Naughton et al. (7) reported coefficients of variation ranging from 4.8 to 9.0% for mean and peak power in the 30-s Wingate test in children aged 6-12 yr when the tests were performed weekly for 8 wk. Similarly, Coggan and Costill (1) observed coefficients of variation of 5.4-6.7% for mean and peak power in 30-s or 60-s tests of maximal cycle ergometry at 90 rpm on four different occasions over 4 wk. Also, Williams et al. (9) reported intraclass correlation coefficients ranging from 0.94 to 0.97 for mean and peak power for two 15-s maximal cycle-ergometer sprints that were separated by either 15 min or 48 h. Finally, Fitzsimmons et al. (3) published the only report to date of the reliability of intermittent anaerobically-biased cycling exercise. Their subjects performed 6 × 6-s maximal-effort cycling sprints on two occasions separated by 6-8 d of rest. Intraclass correlation coefficients for mean power for the individual sprints ranged from 0.90 to 0.97.
None of the studies addressing the reliability of high-power exercise tests, whether intermittent in nature or not, examined the familiarization process in subjects unfamiliar with multiple-sprint tests, i.e., it is unclear how many trials are required before intermittent, high-intensity cycling tests are sufficiently reproducible. Furthermore, it is unknown whether adequate reliability, once established, can be sustained for several nonexercise days. Therefore, the purposes of this study were: 1) to determine the number of trials required to establish high reliability of an intermittent high-intensity cycling test in subjects unfamiliar with multiplesprint performance tests and 2) to determine whether this high reliability can be sustained for a 6-d period.
Five physically fit men participated in the study. Written informed consent was obtained in compliance with the guidelines of the Biomedical Sciences Human Subjects Review Committee for The Ohio State University. The means (± SE) for the subjects' age, height, and body mass were 21.6 ± 1.2 yr, 179.0 ± 1.1 cm, and 79.5 ± 2.6 kg, respectively.
The exercise protocol required the subjects to perform a total of six intermittent, high-intensity exercise test protocols over a 12-d period (Fig. 1). Each subject performed a series of multiple cycling sprints on each of four consecutive days (days 1-4), then rested for 6 d, and finally performed two additional tests on consecutive days (tests 11 and 12). The first four performance tests were designed to establish the number of trials required to establish the extent to which reliability, i.e., consistency in performance, could be achieved within 4 d, whereas tests on days 11 and 12 were conducted to determine if this reliability could be sustained for 6 d.
Subjects did not exercise for 24 h before nor eat within 3 h before each performance test. Before each test, the subject warmed up for 60 s at 50 rpm on a standard friction-braked Monark ergometer set at 0.5-kp resistance. During the next 60 s, the subject mounted a Monark ergometer modified as described by Williams et al. (9) in preparation for initiation of the performance test. After a 6-s countdown, the subject pedaled maximally for 7 s against a fixed resistance (11.34 kg) that replaced the standard adjustable pendular mass on the ergometer. Flywheel rotations were counted to the nearest 1/32 rotation every second during the 7-s sprint (SMI Opto-sensor, Model 1000, 1991 Sports Medicine Industries, Inc., St. Cloud, MN). For each performance test, the 7-s sprint was repeated 10 times, with 30-s rest intervals between sprints. To minimize any possible effect of a subject's anticipation of the end of each exercise bout, the last second of each 7-s sprint was ignored for purposes of statistical analyses.
Peak power output during the first second of the second sprint for each test was selected as the markers of peak power because the coefficients of variation for these data were less than 4% for test days 4, 11, and 12. In contrast, the coefficients of variation for peak power outputs during the first second of the first sprint on each of these days ranged between 10.2 and 14.5%. For statistical comparisons, mean power outputs during sprints 8, 9, and 10 (MP8-10) on each test day were calculated for each of the 4th, 5th, and 6th seconds, i.e., MP8-104th, MP8-105th, and MP8-106th.
Data for mean power and peak power were analyzed with a one-way analysis of variance procedure with repeated measures on time (P ≤ 0.05). When a significant effect of time was observed, Neuman-Keuls multiple-comparison post-hoc tests were used to determine which pairs of means were different.
A comparison of results on test days 1, 2, 3, and 4 determined the extent to which performance varied over these four consecutive daily trials in subjects unfamiliar with multiple-sprint performance tests. For days 3 and 4, values for MP8-104th, MP8-105th, and MP8-106th were greater than on day 1 (Fig. 2; P < 0.05). MP8-106th on day 2 was also greater than on day 1 (Fig. 2; P < 0.05). There were no differences in MP8-10 among days 2, 3, 4, 11, and 12 (Fig. 2). Also, peak power on day 1 was lower (P < 0.05) than peak power for all other days, which were not different from one another (Fig. 3).
The greatest coefficient of variation for MP8-10 was 5.2% for MP8-106th on days 2 versus 3 and on days 2 versus 4. Analysis of data from test days 3 and 4 provides information about the reliability of test performance once the subjects had become familiarized with the protocol, i.e., by day 3, when mean power output had become relatively stable. The coefficients of variation for MP8-104th, MP8-105th, and MP8-106th on test day 3 versus day 4 were 3.3%, 2.5%, and 2.9%, respectively. Thus, once familiarity with the test had been achieved, i.e., apparently by test day 3, the coefficients of variation ranged between 2.5 and 3.3%.
Analyzing the variability in mean power between test days 4 and 11 assessed the persistence of the familiarization effect apparently achieved by day 3. The coefficients of variation for MP8-104th, MP8-105th, and MP8-106th for test-retest comparisons among days 4, 11, and 12 ranged from 2.1 to 3.9%, with an overall mean of 3.1%. The mean coefficient of variation for peak power for all pairwise combinations of days 4, 11, and 12 was 2.8%. Thus, the relatively high reproducibility achieved by test familiarization at the third and fourth test days persisted throughout the 6-d rest period.
The first purpose of this study was to determine the number of trials on consecutive days required to establish high reliability of an intermittent high-intensity cycling test in subjects unfamiliar with tests involving multiple sprints. For the test protocol that was utilized, the mean power outputs achieved during the 4th, 5th, and 6th seconds of sprints 8-10 on the first test day were less (P < 0.05) than those on days 3 and 4. Furthermore, the peak power output on the first test day was less (P < 0.05) than that on test days 2, 3, and 4. Thus, tests 3 and 4 seemed to minimize the variability of both mean and peak power outputs for this test protocol, suggesting that at least two familiarization trials are required before subjects unfamiliar with multiple-sprint tests can achieve power outputs with a reproducibility of 2.5-3.3% in such a test of intermittent high-intensity cycling performance.
We speculate that the improvement in performance from day 1 to days 3 and 4 observed in the present study was due to neural adaptations because the effects were so rapid (6). However, it is also possible that the exercise stimulus on days 1 and 2 increased the activities of glycolytic enzymes and thereby enhanced the glycolytic provision of energy for exercise. Consistent with this hypothesis are reports of concurrent increases in the activities of glycolytic enzymes and in exercise performance after high-intensity interval training (2,4,5,8). However, the duration of these training periods was 5-9 wk, so it is unclear if such changes would occur as a consequence of performing two trials of the exercise test used in the present study.
The second purpose of the present study was to determine whether high reliability of an intermittent high-intensity cycling test could be maintained for 6 d. Because there were no significant differences in mean power and peak power among trials 4, 11, and 12 and because the coefficients of variation for mean and peak power for any combination of these trials were only 2.1-3.9%, it appears that high reliability of an intermittent high-intensity cycling test can be sustained when there is a 6-d period of no exercise, provided there have been four familiarization trials before the 6 d of rest. These findings suggest that the neural or biochemical adaptation(s) to the exercise stimulus on days 1 and 2 persisted for 6 d.
The present study has both evaluated the familiarization to a high-intensity intermittent exercise test protocol in subjects unfamiliar with multiple-sprint performance tests and assessed the reliability of the test protocol once familiarization occurred. We conclude that at least two performance trials on consecutive days should be conducted to ensure reproducible power outputs during high-intensity intermittent cycling exercise. Furthermore, the reliability of such tests is likely to persist for at least 6 d.
1. Coggan, A. R., and D. L. Costill. Biological and technological variability of three anaerobic ergometer tests. Int. J. Sports Med.
2. Costill, D. L., E. F. Coyle, W. F. Fink, G. R. Lesmes, and F. A. Witzmann. Adaptations in skeletal muscle following strength training. J. Appl. Physiol.
3. Fitzsimmons, M., B. Dawson, D. Ware, and A. Wilkinson. Cycling and running tests of repeated sprint ability. Aust. J. Sci. Med. Sport.
4. MacDougall, J. D., A. Hicks, J. R. Macdonald, R. S. McKelvie, H. Green, and K. M. Smith. Muscle performance and enzymatic adaptations to sprint interval training. J. Appl. Physiol.
5. Linossier, M. T., D. Dormois, C. Perier, J. Frey, A. Geyssant, and C. Denis. Enzyme adaptations of human skeletal muscle during bicycle short-sprint training and detraining. Acta Physiol. Scand.
6. Moritani, H., and H. A. Devries. Neural factors versus hypertrophy in the time course of muscle strength gain. Am. J. Phys. Med.
7. Naughton, G., J. Carlson, and I. Fairweather. Determining the variability of performance on Wingate anaerobic tests in children aged 6-12 years. Int. J. Sports Med.
8. Roberts, A. D., R. Billeter, and H. Howald. Anaerobic muscle enzyme changes after interval training. Int. J. Sports Med.
9. Williams, J., W. Barnes, and J. Signorile. A constant-load ergometer for measuring peak power output and fatigue. J. Appl. Physiol.