Coach Tips

The effect of manipulating high-intensity interval training programs by work-bout duration and intensity on time-trial performance in cyclists:
​a remote randomized trial.

Published on October 14, 2022
​
Updated on October 21, 2022

Michael Rosenblat
Jem Arnold
​Scott Thomas

KEY POINTS

1. Virtual training platforms can provide coaches, exercise physiologists, and researchers a reliable method to program and monitor exercise testing and training sessions for athletes
   
2. Remote training studies should include direct verbal communication with applicants during the recruitment process to ensure that all participants understand the study requirements and subsequently maintain higher adherence rates. 
   
3. Exercise programming that accounts for differences in submaximal performance allows for a more individualized approach to exercise prescription. 

ABSTRACT

The primary objective of this study was to evaluate the effects of high-intensity interval training (HIIT) programs that differed in either work-bout duration or work-bout intensity on time-trial (TT) performance in trained cyclists. Participants were recreational to elite male (n = 35) and female (n = 5) cyclists (age = 35.3 ± 7.0 years, body mass = 73.1 ± 6.6 kg, Wpeak = 4.8 ± 0.7 W×kg-1, WTT = 3.4 ± 0.6 W×kg-1) who were recruited and performed all training sessions remotely and unsupervised using a commercially available virtual cycle training platform. They were randomly assigned to one of four HIIT groups that varied in work-bout duration and intensity. Participants were asked to complete two HIIT sessions per week for 6-weeks, with additional low intensity continuous training at a prescribed workload, all performed on the virtual training platform. The results indicated that intervals with 6-minute work-bouts led to a greater improvement in Wpeak compared to the other groups. However, when accounting for previous interval training experience, there was no difference among the groups. There was a significant correlation between baseline Wpeak (W×kg-1) with changes in Wpeak (r = -0.38, p < 0.01) and changes in WTT (r = -0.48, p < 0.01). Participant dropout rate was high across all groups, with only 23% of initially recruited subjects completing the full 6-week training program. The lack of statistical significance among group differences in Wpeak and WTT post intervention may have been due to the high number of participant dropouts.

1    INTRODUCTION

Exercise programming designed to optimize endurance performance improvements is complex, with many interventions that have been shown to be effective. Interval training in particular, has become a popular modality since it can produce greater improvements in maximal oxygen consumption (V̇O2max) than continuous training, and in a relatively short period of time [1]. High-intensity interval training (HIIT) consists of repeated bouts of exercise within the severe intensity domain (i.e., at an intensity above the maximum metabolic steady state; e.g., critical power (CP) or maximum lactate steady state (MLSS)) but below the highest power or speed that allows for the attainment of  V̇O2max) [2]. Sprint interval training (SIT) occurs within the extreme intensity domain (i.e., at an intensity that coincides with a power or speed above V̇O2max) [2]. Both HIIT and SIT can produce similar improvements in V̇O2max. However, HIIT has been shown to be the superior mode of interval training for improving time-trial (TT) performance [2].

        The results of a meta-analysis on programming interval training indicate that improvements in TT performance following HIIT are dependent on a participant’s training status, with untrained (inactive or active) individuals improving to a greater extent than trained individuals [3]. In this case, trained individuals were those who engaged in a structured training program specific to a mode of exercise [3]. While these results may be intuitive, the study found that trained individuals may have a greater response to a more individualized interval training program. Specifically, HIIT programs with longer duration work-bouts were shown to lead to greater improvements in TT performance. While this meta-analysis was able to provide insight into the relationship between training characteristics with changes in TT performance, the results did not include pairwise analyses of studies that directly compared different HIIT programs.

        There are several primary research investigations that have directly assessed the influence of different HIIT programs on measures including V̇O2max, peak power (Wpeak) from an incremental exercise test (IET) and TT performance [4-6]. However, there are conflicting results among the studies, which are likely be due to the heterogeneity in their methodologies and populations. A common issue in their design is that there are differences in the total external work completed among the HIIT groups. Currently, there are no studies that compare the effects of HIIT exercise differing in work-bout duration but with identical external work (i.e., same intensity and total time).

        To further complicate interval programming, a modified form of HIIT that includes work-bouts that consist of intermittent exercise (e.g., 30-sec work : 15-sec recovery) has been shown to be superior to HIIT with continuous work-bouts [7, 8]. It is noteworthy that studies that have investigated the acute response to this intermittent approach have shown that it allows for participants to spend a greater amount of time above 90%V̇O2max when compared to a more traditional HIIT program [9, 10]. Exercising at a high percentage of V̇O2max has been suggested to be one of the mechanisms leading to adaptations in performance. However, a causal relationship between time at V̇O2max with changes in V̇O2max or TT have not been identified. Nevertheless, there are also no studies that compare an intermittent interval program with a longer duration HIIT program that maintains identical workload among groups. It would be beneficial to determine the influence of interval work-bout duration on TT performance when total mechanical work is matched.

        Many interval training studies require participants to visit a laboratory to use expensive equipment for data collection purposes. This can lead to a several issues that can negatively influence training studies. For example, due to time constraints, location, and various other reasons it can be difficult to recruit individuals to participate in on-site studies. This has led to a high prevalence of sport science studies with small sample sizes [11]. In addition, it is often the case that there are limitations to the practical implementation of sport science research at both the practitioner and coaching level. Practitioners have reported that the lack of applied research that can be directly implemented into practice is one of the missing links between sport science research and its translation into practice [12]. Today there are several virtual training and racing platforms available online that athletes can use right from their homes and all times of the year [13]. These platforms offer structured training plans and the ability to design custom workouts and training programs in lieu of personalised coaching. Sport science, as an academic discipline, would benefit from considering how experimental efficacy can be translated to real-world effectiveness in virtual training environments where endurance training programs are commonly executed [13].

​        The primary objective of this study was to evaluate the effects of HIIT programs that differed in either work-bout duration or work-bout intensity on TT performance in trained cyclists. We hypothesized that longer duration HIIT-bouts would lead to greater improvements in performance. The second objective of this study was to determine the feasibility of a remote and virtually delivered method to program HIIT using a commercially availabe online training platform. We hypothesized that this would increase reruitment and adherence rates, and provide meaningful and practical results. We also can see this  methodology adapted for future studies and directly implemented into practice.

2    METHODS

2.1   Study Design

The study was approved by the University of Toronto Research Ethics Board, Toronto, Ontario, Canada. A minimum of 140 participants were to be recruited, assuming 90% power, an alpha of 0.0167 adjusted for multi-arm comparison, and up to 20% attrition per group. Participants were randomly assigned to one of four HIIT groups with an allocation ratio of 1:1:1:1 using a computer-generated random sequence program with random block sizes. This process allowed for concealed allocation of participants to their respective groups. Participants were blinded to the HIIT programs used in the other groups. All participants were provided with a coded TrainerRoad (TrainerRoad LLC, Reno, Nevada, USA) account to record all testing and training data, which ensured blinding of assessors. TrainerRoad is a commercially available virtual training platform that can be used for planning and performing indoor cycling training.

​        Testing was conducted for the two weeks prior to and one week following a 6-week HIIT intervention. The first week of baseline testing included an incremental exercise test (IET) and an initial TT test for familiarization. In the second week, participants performed a second TT test which was used as their baseline measurement. All testing was completed with at least 48-hours between tests. Once baseline testing was completed, participants performed their respective HIIT programs which consisted of two HIIT sessions per week. Participants were allowed to complete as many low-intensity continuous training (LICT) sessions in addition to their respective HIIT interventions, as long as those sessions were performed using the TrainerRoad software platform. See Figure 1 (click here) for the timeline for the testing and training, and daily exercise schedule.

        All testing and training sessions were completed by participants at their residence on their own bicycles with the use of electronically controlled ‘smart’ trainers that were compatible with the TrainerRoad platform. It was recommended that all testing and training be conducted in the same room and time of day with the use of a fan to regulate ambient temperature. Heart rate measurements were taken via each participant’s heart rate monitor. Data for cycling cadence was collected for those participants who had the appropriate sensors. The data was collected using the TrainerRoad software platform which was installed on the participants’ own computer or smart phone devices.

​        Participants were asked to keep a food diary for the day leading up to each of the testing sessions. The diary was used to replicate their diet for the following testing weeks. Participants were asked to refrain from engaging in strenuous exercise for at least 48 hours, from consuming alcohol for at least 24 hours, and from consuming caffeine for at least 2 hours prior to all test sessions. In addition, they were not to consume a large meal within 2 hours of each of the tests.

2.2   Participants

Participants were recruited by using online forums, podcasts, emails to cycling clubs and academic varsity programs, and word-of-mouth. The investigator also posted notices on endurance sport online forums and reached out to the hosts of relevant podcasts and online sport news outlets to further the reach of the recruitment.

​        The experimental procedures and risks associated with participating in the study were outlined in writing through a web survey using the REDCap (Vanderbilt University, TEN, USA) data management web application. All individuals provided written informed consent prior to taking part in the study. In addition to providing consent, they were asked to complete a training history that provided a general outline of the previous training experience. This data was also collected using REDCap.

        Participants were included if they were between the ages of 18 and 45 years of age and have been cycling for a minimum of 3 times per week for the previous 6 months. Participants were excluded from the study if they had any injury in the previous 6 months that prevented them from completing their normal training routine, or if they had any health impairments including cardiovascular diseases (e.g., heart attack, hypertension), neurological conditions (e.g., stroke, peripheral neuropathy) or metabolic conditions (e.g., diabetes).

2.3   Measurements

2.3.1 Incremental Exercise Test

Participants performed a 10-minute warm up at a power output of 100 watts. Following their warm up, participants performed a zero offset to their trainer according to manufacture instructions, then completed an additional 5-minutes at 100 watts. The first stage of the IET was completed at 100 watts and then workload was increased by 12.5 watts every 30-seconds until the participant was unable to maintain a cadence of 70 rpm. Wpeak was recorded as the final completed 30-second stage.

2.3.2 Time-Trial Test

The TT test began with a 10-minute warm up at 100 watts. Participants then performed a trainer zero offset and performed 3-minutes at 100 watts, followed by 2-minutes at a self-selected intensity that they believed they could maintain for the 40-minute TT test. The participants then returned to 100 watts for the final 5 minutes of the warm up. The 40-minute TT was performed at their best average power.

2.4   Intervention

All training sessions commenced with a 10-minute warm up at an power output equal to 40% Wpeak determined from the IET. Participants then performed a trainer zero offset and completed an additional 5 minutes at the same intensity. There were four different HIIT groups which differed by either work-bout duration or by work-bout intensity.

        The HIIT programs were pre-programmed into the TrainerRoad virtual training platform to ensure that participants completed the programs accurately. Briefly, groups consisted of the following: Group 1 (HIITLong), four repetitions of 6-minute work-bouts at 15% of the power difference between the TT mean power (WTT) and Wpeak; Group 2 (HIITSHORT-LOW), twelve repetitions of 2-minute work-bouts at the same intensity as Group 1; Group 3 (HIITSHORT-HI), the same number of sets and repetitions as Group 2 but at 30% of the power difference between WTT and Wpeak; and Group 4 (HIITINT), which included an intermittent protocol at the same intensity as Group 3. See Table 1 (click here) for a detailed description of the interval training programs.

        LICT was performed at 40% Wpeak and the duration was self-selected in sessions ranging from 30-mintues to 4-hours. Participants were asked to refrain from performing any additional exercise except their usual activities of daily living (e.g., commuting to work, grocery shopping, etc.).

3    STATISTICAL ANALYSIS

All data are expressed as a mean ± standard deviation (SD) unless otherwise specified. Data analysis was performed using custom scripts written in the computing environment R (v4.1.2, R Foundation for Statistical Computing, Vienna, Austria). Shapiro-Wilk Test was used to assess the normality of the sample. A 1-way ANOVA was used to assess differences among the four groups at baseline and for training program characteristics. A 2-way repeated measures ANOVA was used to assess the interaction effect of group and time on the performance measures. Post-hoc analysis of the performance measures was conducted using Tukey’s Honestly Significant Difference which provides an adjusted p-value to account for the number of comparisons performed. Pearson’s correlation coefficient was used to determine the relationship among baseline and training characteristics for the two performance measures, Wpeak and WTT.

3    RESULTS

A total of 171 (20 female) participants were recruited to participate in the study. Only 40 (23%) of registered participants completed follow up testing. Four participants were removed due to randomization error, 12 dropped out due to illness or injury, one dropped out due to psychological fatigue, 9 dropped out because they wanted to perform a different training program, and 98 participants did not provide a reason for dropping out. Figure 2 (click here) provides the details of the participant randomization process and completion rate. Only participants who completed the study were included in the results.

        The groups were similar at baseline for both Wpeak and WTT. See Table 2 (click here) for a full description of the group characteristics.

        Table 3 (click here) includes the results of the training completed by the different groups. The interval work-bout intensity was significantly higher for HIITSHORT-HI and HIITINT compared to HIITLONG and HITSHORT-LOW, as designed. There were no other differences in training programs among the 4 groups.

        There was a significant (p < 0.05) increase in Wpeak from baseline for groups 1, 3 and 4 (Table 4 (click here)). There were no within group changes in any other variables from baseline. In addition, there were no between group differences among any of the groups for Wpeak or WTT at follow up. There was a significantly greater change from baseline in Wpeak for Group 1 compared to the other groups (Table 5 (click here)). However, two of the participants in Group 1 did not have previous interval training experience. When accounting for these outliers, there was no longer a significant difference among the groups.

        There was a significant correlation between baseline Wpeak (W×kg-1) with changes in Wpeak (r = -0.38, p < 0.01) and changes in WTT (r = -0.48, p < 0.01). There was no correlation between any of the other participant characteristics with change in the performance measures.

4    DISCUSSION

This is the first study to our knowledge that has utilized a commercially available virtual training platform to conduct baseline and follow up performance testing, and to administer the different intervention programs. Due to time constraints, location, and other logistical limitations it can be difficult to recruit individuals to participate in on-site studies. This has led to a high prevalence of sport science studies with small sample sizes [11]. The COVID-19 pandemic accelerated the trend of athletes using virtual training software to run their own exercise sessions at home. At the same time, it put a temporary halt on the ability to perform laboratory data collection  [13]. It will be important to consider the strengths and limitations of remote-based virtual training for sport science studies [13]. Furthermore, it is also necessary to evaluate the validity and reproducibility of exercise tests performed under unsupervised, remote and home-based conditions [14, 15]. We believe these trends in home-based virtual training and remote data collection will continue to expand, and important lessons can be learned from the early attempts, such as the present study. We hypothesized that conducting a study with the use of a virtual training platform would allow for a larger and more diverse sample group and greater adherence rates compared to studies of similar design.

        Previous findings suggest that women make up only 39% of participant in publications in sport and exercise journals [16]. We hoped that the remote nature of the study would allow for a greater percentage of female participants. In addition, to further recruit female participants we directly communicated with female cyclists to share study recruitment information with their team/club members. However, even with these strategies only 12% of the participants who were recruited were female, and only 13% of participants who completed the study were female.

        We recruited 171 participants over a 6-month period. Due to the large sample size, we did not believe it was necessary to stratify participants prior to randomization. This is a common practice to ensure baseline similarity in sport science studies as they often include smaller sample sizes. However, only 23% (40 of 171) of participants who were recruited and randomized completed the study. Group 2 (HIITSHORT-LOW) had the highest number of dropouts; however, there was no significant difference in characteristics among those participants who left the study. While the high number of participant dropouts can influence group characteristics, undermining the randomization process, in this case there was no difference in mean values among the groups at baseline. The inability to detect differences among the groups may be because the standard deviations for each of the baseline values were large, indicating that the participant characteristics within the respective groups were highly heterogenous. Therefore, the inability to determine a difference at baseline among the groups may be due to regression to the mean, causing a type II error (false negative).

        Approximately 50% of the participants who initially volunteered and were randomized into their respective groups did not begin testing or respond to the investigators regarding their reason for dropping out. Of those participants who did respond, many reported that they wanted to perform training that was not consistent with the 6-week intervention program, including strength training and unstructured group rides. The study requirements were explained in writing in the recruitment advertisements and in the written informed consent document. A higher expected dropout rate may need to be considered for future remote sport science studies. In addition, it would be beneficial to verbally communicate the study requirements to each individual and to give additional opportunities for participants to ask questions prior to recruitment or randomization. This may lead to greater adherence.

        To the best of our knowledge, this was the first study that investigated the effect of HIIT programs that differed by interval work-bout duration but consisted of identical external workload. We hypothesized that the interval training protocol with longer duration intervals (i.e., those with 6-min work-bouts) would lead to greater improvements in WTT and Wpeak compared to the protocol with shorter work-bouts (2-min work-bouts). However, the results of the current study did not show a significant difference between the two intervention groups, likely due to the high number of participant dropouts.

        Previous studies have shown that longer duration intervals can produce greater improvements in Wpeak and V̇O2max [5, 6] and TT performance [5]. A limitation to these studies is that the training protocols did not include identical total external work per session. Furthermore, the groups differed by more than one programming variable (work-bout intensity and work-bout duration). The results of our previous meta-analysis showed that the total duration of work completed per session did not influence change in TT performance [3]. However, these findings were through a meta-regression, not a direct comparison between groups, as was done in the current study.

        An interesting finding from the meta-analysis was that HIIT work-bout intensity did not influence changes in TT performance. There are several studies that directly compared the effect of manipulating interval intensity on changes in TT performance [5, 17-19], which showed conflicting results. Similar to the limitation in the studies on interval work-bout duration, the intervention programs in these studies differed by more than one programming variable (i.e., both intensity and duration). Therefore, we included a third group to compare interval programs that only differed by intensity (HIITLOW and HIITHI). Based on the results of our meta-analysis, we hypothesized that there would be no differences between HIITLOWand HIITHI as long as all subjects were performing work bouts in the severe intensity domain. This was which was consistent with the results of the current study. As previously mentioned, there were a high number of participant dropouts, which likely influenced the results.

        Time spent above 90% of V̇O2max is one hypothesis regarding the mechanism leading adaptations in endurance performance [20, 21]. It is important to note that a causal relationship between time at V̇O2max with changes in V̇O2max or TT have not been identified. Due to a link between the metabolic demand at these intensities and theory of specificity, exercise physiologists commonly utilize interval training programs designed to maximize time near V̇O2max. Rønnestad et al. [9] found that shorter intervals (30-sec work-bouts) with a work-to-rest ratio of 2:1 may lead to a greater time spent near V̇O2max. They also found that 10 weeks of effort matched intermittent intervals (30-sec work : 15-sec recovery) led to a greater improvement in 40-min TT performance compared to longer duration intervals that consisted of 5-min work-bouts and 2.5-min recovery periods [8]. Since they programmed exercise by best effort, the two protocols were performed at different intensities and completed different total work (kJ). Therefore, in the current study we included an intermittent interval group (HIITINT) that only differed from the HIITHI group by the distribution of the work and rest bouts. Again, due to the high number of participant dropouts, the results did not show a difference in Wpeak or WTTamong the groups at follow up.

        We used a novel approach to programming exercise intensity to account for individual differences in submaximal performance. It is common practice to program high-intensity interval exercise by using a percentage of Wpeak that coincides with a V̇O2 value close to peak levels. However, this method may result in participants training at different relative intensities, or even in different intensity domains entirely, as it does not account for individual differences in the metabolic steady-state threshold as a fraction of V̇O2max. Furthermore, there are several different protocols that can be used to determine Wpeak. Protocols that are longer than 12-minutes (increment stages greater than 1-minute) can result a lower Wpeak [22], making it difficult to compare exercise intensity among studies.

        The IET in the current study incorporated shorter stage increments (12.5 watt increases every 30-seconds) to ensure that total test time was under 12 minutes. Interval work-bout intensity was programmed at a percentage of the difference between Wpeak and the WTT (15% for groups 1 and 2, and 30% for group 3 and 4); where WTT was used as a proxy for the metabolic steady-state threshold (i.e., CP). We used these values because the results of our pilot testing (unpublished) allowed for the greatest session completion rate in a group of cyclists with various fitness levels. Most of the cyclists who piloted the interventions were unable to complete the HIITLONG protocol when intensity was above 15% of the difference between Wpeak and WTT. In addition, most of the cyclists were unable to complete the HIITSHORT-HI or the HIITINT protocols when intensity was above 30% of the difference between Wpeak and WTT. In the current study, the group mean power output for WTT was 70% ± 4% of Wpeak. This is consistent with CP and MLSS as shown in previous literature where CP can occur at 67% of Wpeak [23] and MLSS at 70% of Wpeak [24]. The mean power output during the interval sessions for all subjects in the current study was 77% ± 4% of Wpeak (75% ± 3% for groups 1 and 2, and 79% ± 3% for groups 3 and 4). Therefore, it is highly likely that all participants were completing their exercise sessions within the severe intensity domain, as intended.

​        We believe the lessons learned from this study lead to some practical advice for researchers performing remote training intervention studies. As mentioned, the starting expectation for participant adherence may be considerably lower than for on-site laboratory research, however we believe this can be reduced with a few simple, proactive measures. First, despite the promise of scaling up sample sizes in remote sport science research, direct personal (video or voice) communication with each subject may be necessary to proactively encourage continued adherence. Many of the subjects who dropped out disengaged with, or were not aware in the first place, of the available communication channels with the researchers. If they had been proactively contacted soon after recruitment, they may have been more incentivized to begin and continue the study. Second, the time of season for recruitment of competitive subjects should be considered, to minimize interruption of existing training plans and competitive obligations. We began recruitment in November, 2021 during the northern hemisphere off-season, however already at this time many athletes are committed to and beginning their training programmes for the upcoming season. Earlier recruitment at the end of the summer competitive season and allowing subjects to delay the start of their testing and training phase on their own schedules, may encourage higher engagement. Finally, we did not have equal representation of female and male athletes in our recruited or completed sample group. We feel this can easily be a self-imposed requirement of remote training research, where recruitment can be targeted toward the under-represented groups until an approximately equal distribution is reached.

5    CONCLUSION

The results indicated that intervals with 6-minute work-bouts led to a greater improvement in Wpeak compared to the other groups. However, when accounting for previous interval training experience, there was no longer a difference among the groups. The lack of statistical significance in group differences in Wpeak and WTT post intervention may be due to the high number of participant dropouts. Future remote training studies should include direct verbal communication with applicants during the recruitment process to ensure that all participants understand the study requirements and subsequently maintain higher adherence rates.

AUTHOR CONTRIBUTIONS

Michael Rosenblat created the project, ethics submission, study design, recruitment, data collection, stakeholder engagement, statistical analysis, manuscript preparation. Jem Arnold participated in the study design, pilot testing of the training protocols, recruitment, data collection, stakeholder engagement, manuscript preparation. Scott Thomas participated in Study design, ethics submission, protocol submission, manuscript preparation. All authors read and approved the final manuscript.

CONFLICT OF INTEREST

Michael Rosenblat, Jem Arnold and Scott Thomas declare that they have no conflicts of interest relevant to the content of this study.

FUNDING

No sources of funding were used to assist in the preparation of this article.

ACKNOWLEDGEMENTS

First, we would like to thank all the participants for their participation in the study. The authors would like to thank the following individuals for their help and support with study: Nate Pearson, the CEO and Co-Founder of TrainerRoad for providing access to the TrainerRoad software and full support from the TrainerRoad team; Corey Croasdell, the Customer Support Manager at TrainerRoad for his invaluable support throughout the data collection process; Ryan Lyn who helped in the development of the REDCap registration form; Bruno da Costa who consulted with the study design and participant randomization process; Bent Rønnestad, who consulted with the study design and participant recruitment; Stephen Seiler, who consulted with the study design and participant recruitment; and Todd Astrino, who consulted with the study design and participant recruitment.

REFERENCES

1. Milanović Z, Sporis G, Weston M. Effectiveness of high-intensity interval training (HIT) and continuous endurance training for VO2max improvements: a systematic review and meta-analysis of controlled trials. Sports Med. 2015;45(10):1469-81.
   
2. Rosenblat MA, Perrotta AS, Thomas SG. Effect of high-intensity interval training versus sprint interval training on time-trial performance: a systematic review and meta-analysis. Sports Med. 2020;50(6):1145-61.
   
3. Rosenblat MA, Lin E, da Costa BR, Thomas SG. Programming interval training to optimize time-trial performance: a systematic review and meta-analysis. Sports Med. 2021;51(8):1687-714.
   
4. Helgerud J, Høydal K, Wang E, Karlsen T, Berg P, Bjerkaas M, et al. Aerobic high-intensity intervals improve VO2max more than moderate training. Med Sci Sports Exerc. 2007;39(4):665-71.
   
5. Stepto NK, Hawley JA, Dennis SC, Hopkins WG. Effects of different interval-training programs on cycling time-trial performance. Med Sci Sports Exerc. 1999;31(5):736-41.
   
6. Seiler S, Jøranson K, Olesen BV, Hetlelid KJ. Adaptations to aerobic training: interactive effects of exercise intensity and total work duration. Scand J Med Sci Sports. 2013;23(1):74-83.
   
7. Rønnestad BR, Hansen J, Nygaard H, Lundby C. Superior performance improvements in elite cyclists following short-interval vs effort-matched long-interval training. Scand J Med Sci Sports. 2020;30(5):849-57.
   
8. Rønnestad BR, Hansen J, Vegge G, Tonnessen E, Slettalokken G. Short intervals induce superior training adaptations compared with long intervals in cyclists - an effort-matched approach. Scand J Med Sci Sports. 2015;25(2):143-51.
   
9. Rønnestad BR, Hansen J. Optimizing interval training at power output associated with peak oxygen uptake in well-trained cyclists. J Strength Cond Res. 2016;30(4):999-1006.
   
10. Almquist NW, Nygaard H, Vegge G, Hammarström D, Ellefsen S, Rønnestad BR. Systemic and muscular responses to effort-matched short intervals and long intervals in elite cyclists. Scand J Med Sci Sports. 2020;30(7):1140-50.
    
11. Heneghan C, Perera R, Nunan D, Mahtani K, Gill P. Forty years of sports performance research and little insight gained. BMJ. 2012;345:e4797.
    
12. Reade I, Rodgers W, Spriggs K. New ideas for high performance coaches: a case study of knowledge transfer in sport science. Int J Sports Sci Coach. 2008;3(3):335-54.
    
13. McIlroy B, Passfield L, Holmberg HC, Sperlich B. Virtual training of endurance cycling - a summary of strengths, weaknesses, opportunities and threats. Front Sports Act Living. 2021;3:631101.
    
14. Matta G, Edwards A, Roelands B, Hettinga F, Hurst P. Reproducibility of 20-min time-trial performance on a virtual cycling platform. Int J Sports Med. 2022;Online ahead of print.
    
15. Maunder E, Rothschild JA, Ramonas A, Delcourt M, Kilding AE. A three-minute all-out test performed in a remote setting does not provide a valid estimate of the maximum metabolic steady state. Eur J Appl Physiol. 2022;122(11):2385-92.
    
16. Costello JT, Bieuzen F, Bleakley CM. Where are all the female participants in Sports and Exercise Medicine research? Eur J Sport Sci. 2014;14(8):847-51.
    
17. Denadai BS, Ortiz MJ, Greco CC, de Mello MT. Interval training at 95% and 100% of the velocity at VO2max: effects on aerobic physiological indexes and running performance. Appl Physiol Nutr Metab. 2006;31(6):737-43.
    
18. Smith TP, Coombes JS, Geraghty DP. Optimising high-intensity treadmill training using the running speed at maximal O(2) uptake and the time for which this can be maintained. Eur J Appl Physiol. 2003;89(3-4):337-43.
    
19. Turnes T, de Aguiar RA, de Oliveira Cruz RS, Pereira K, Salvador AF, Caputo F. High-intensity interval training in the boundaries of the severe domain: effects of sprint and endurance performance. Int J Sports Med. 2016;37(12):944-51.
    
20. Billat LV, Koralsztein JP. Significance of the velocity at VO2max and time to exhaustion at this velocity. Sports Med. 1996;22(2):90-108.
    
21. Wenger HA, Bell GJ. The interactions of intensity, frequency and duration of exercise training in altering cardiorespiratory fitness. Sports Med. 1986;3(5):346-56.
    
22. Amann M, Subudhi A, Foster C. Influence of testing protocol on ventilatory thresholds and cycling performance. Med Sci Sports Exerc. 2004;36(4):613-22.
    
23. Brickley G, Doust J, Williams CA. Physiological responses during exercise to exhaustion at critical power. Eur J Appl Physiol. 2002;88(1-2):146-51.
    
24. Dekerle J, Baron B, Dupont L, Vanvelcenaher J, Pelayo P. Maximal lactate steady state, respiratory compensation threshold and critical power. Eur J Appl Physiol. 2003;89(3-4):281-8.