The influence of cadence on fatigue during maximal sprint cycling in world-class and elite sprint cyclists

ABSTRACT

Optimising cadence through appropriate gear selection is a key consideration for track sprint cycling performance, yet the influence of cadence on fatigue (i.e., decrement in power output) within a maximal sprint is not well understood. The aim of this study was to identify the influence of cadence on fatigue during maximal sprint cycling. Eleven world-class and elite track sprint cyclists (n = 6 men, maximal power output (P_max) = 1894 ± 351 W, optimal cadence (F_opt) = 134 ± 8 rev∙min⁻¹: n = 5 women, P_max = 1114 ± 80 W, F_opt = 124 ± 8 rev∙min⁻¹) completed two testing sessions where power-cadence profiles were constructed to determine the F_opt associated with P_max. Cyclists also performed three maximal 15-s sprints (F_opt, ±15%F_opt) to identify fatigue per pedal stroke across these cadence ranges. There was no significant difference (p = 0.2) in the absolute fatigue per pedal stroke when cadence was fixed 15% above (16.7 ± 6.1 W∙stroke⁻¹) and below (15.3 ± 5.1 W∙stroke⁻¹) F_opt. Similarly, there was no significant difference in the relative fatigue per pedal stroke (% peak power∙stroke⁻¹) across F_opt and ± 15%F_opt trials (p = 0.12). The relative decrement in power output is equivalent across the ± 15%F_opt cadence range. As such, a higher-geared, lower-cadence approach to maximal sprint cycling could be a viable method to minimise maximal pedal strokes and reduce the decrement in power output.

KEYWORDS:

• Fatigue
• sprint cycling
• elite
• cadence
• world-class

Introduction

Muscular fatigue is defined as an exercise-induced decrease in the ability to produce force or power, which may arise due to changes at the peripheral or central level (Enoka & Duchateau, 2008; Gandevia, 2001; Wan et al., 2017). Exercise-induced reduction in maximal power output (P_max) is a primary limitation for track sprint cycling events (Douglas et al., 2021), given that P_max is a key determinant of a range of competitive cycling events, including flying-start 200-m (f200-m) cycling performance in elite track cyclists (Dorel et al., 2005). P_max can be achieved under particular mechanical constraints by reaching an optimal cadence (F_opt) through appropriate gear selection. The importance of cadence selection during maximal sprint cycling bouts is demonstrated through the polynomial relationship between power output and cadence (i.e., P-C profile) (Dorel, 2018; Dorel et al., 2005; McCartney et al., 1983; Rudsits et al., 2018; Wackwitz et al., 2023; Wackwitz et al., 2020). Despite the importance of cadence selection on an individual’s ability to produce power, the effect of cadence on the magnitude of power output decline (i.e., fatigue rate) is not well understood. As a result, it is likely that recommendations for gear selection for elite track sprint cyclists are based on unfatigued P-C profiles. As such, identifying fatigue rates during maximal sprint cycling is fundamental to performance prediction and optimal gear selection.

The quantification of fatigue rate requires the determination of a decrement in power or force production as a function of duration, distance, or pedal strokes. There has been limited research investigating the effect of cadence selection on fatigue rate within a single short duration maximal sprint (Beelen & Sargeant, 1991; McCartney et al., 1983), in particular, as a function of cumulative pedal strokes (Tomas et al., 2010). However, studies of longer duration maximal efforts can be used to gain inferences into the effect of cadence on maximal performance, whereby high cadence efforts have been shown to result in higher peak power output (P_peak) and fatigue rate in 3-min all-out test (de Lucas et al., 2014; Wright et al., 2017). An earlier study from McCartney et al. (1983) investigated the effect of cadence on fatigue rate and established that lower cadences resulted in a less severe decrement in power output during a 30-s maximal cycling sprint. This finding was mirrored by Beelen and Sargeant (1991). However, within these studies (Beelen & Sargeant, 1991; McCartney et al., 1983), the P_peak of each effort was not matched. Therefore, it remains unclear whether the fatigue rate at low cadences is less severe than at high cadences due to the slower contraction velocity, a lesser number of cumulative pedal strokes completed, or the higher P_peak at faster pedalling rates, which may have led to an increased accumulation of metabolic by-products associated with increased fatigue (Fitts, 2008; McLester, 1997; Tomas et al., 2010). The results from their investigation also show that the mean power output of the high cadence trial was greater than that of the low cadence trial, despite this trial possessing a significantly greater fatigue rate (McCartney et al., 1983). As such, to optimise power output, there is clearly a balance between selecting a gear that provides the mechanical constraints allowing a cyclist to achieve a cadence range to reduce fatigue, while also maximising power output.

The investigation by Tomas et al. (2010) addressed these limitations by completing two optimised maximal sprint cycling bouts with different crank lengths. These optimised bouts were matched in P_peak. As such, this investigation mitigated previous limitations and established that the number of cumulative maximal pedal strokes completed explained a variance of 98.9% in power output during the two trials. This finding is particularly interesting in its application to elite cycling on the premise that, if the magnitude of fatigue is fixed per maximal pedal stroke, sprint cyclists could prioritise sprinting at a cadence that not only maximises maximal power output but also minimises the number of pedal strokes. A key consideration from these studies (McCartney et al., 1983; Tomas et al., 2010) is that power output was not normalised to the P_peak achieved, which would allow for identification of relative fatigue rates. Both studies (McCartney et al., 1983; Tomas et al., 2010), quantified fatigue index (FI), however, this metric is biased towards greater P_peak. This, however, would not affect the results of Tomas et al. (2010) given the matched P_peak. Given the methodology utilised by Tomas et al. (2010), the results may be confounded as, despite an average optimal cadence being set for each participant based on previous research (Martin & Spirduso, 2001), this was not individualised for each participant based on their own P-C profile, which has previously been shown to be variable between individuals (Wackwitz et al., 2020). In addition, altering the crank length as within the Tomas et al. (2010) investigation also alters the muscle length, which has been demonstrated to induce different metabolic and neuromuscular responses during exercise (Pethick et al., 2021).

As such, the aim of this investigation was to identify if cadence affects the absolute and relative rate of fatigue during maximal sprint cycling and to determine if the difference in fatigue between trials is explained by the cumulative pedal strokes completed. We hypothesise that there will be no significant difference in the absolute fatigue rate expressed as a function of maximal pedal strokes when P_peak matched trials are completed.

Materials and methods

Participants

Five female (20.0 ± 2 years, 70.0 ± 6.0 kg) and six male (21.0 ± 2 years, 88.2 ± 8.2 kg) world-class or elite track sprint cyclists, free from known injury or illness, volunteered for the present study (McKay et al., 2022). The f200-m personal best time for male cyclists (n = 6) was between 9.53 and 9.975 s and 10.64 to 11.41 s for female cyclists (n = 5). All participants had competed at international-level track cycling events. Three participants were classified as world-class athletes with podium places at major international track-sprint events, with the remaining nine participants being classified as elite. Due to COVID-related interruptions, one participant (not included within n = 11) was unable to complete testing sessions within the required timeline and as such, data were excluded from the study. Prior to participation, all participants provided written informed consent according to the Declaration of Helsinki, and it was approved by the Griffith University Human Research Ethics Committee.

Design

This project utilised a within-subject design in which participants attended two laboratory testing sessions at the Queensland Academy of Sport laboratory (Nathan, Queensland, Australia) or Adelaide Superdrome (Gepps Cross, South Australia, Australia), separated by a minimum of 2 days.

Methodology

Lode (Lode Excalibur Sport PFM, Lode B. V., Groningen, the Netherlands) and a custom motor-driven SRM ergometer were utilised for the laboratory testing sessions, whereby sessions completed at the Queensland Academy of Sport laboratory utilised the Lode ergometer and sessions completed at Adelaide Superdrome utilised the SRM ergometer. Both ergometers can function in an isokinetic mode (Wackwitz et al., 2020), whereby the set cadence is maintained throughout the effort and the participant’s ability to produce force determines the power output. The ergometer was set up for each participant based on their competition bicycle setup. A standard crank length of 170 mm was set, and participants used their own cleated shoes with pedals kept constant between participants.

Sessions

Two testing sessions were completed to profile the athletes’ ability to produce power output and investigate the decrement in power output during maximal sprints.

Session 1

Session 1 comprised three seated, 5-s maximal sprints, and a maximal 15-s seated sprint, completed on a stationary cycling ergometer in the laboratory: LODE Excalibur sport PFM or custom SRM motor-driven ergometer. The initial block of 5-s maximal sprints was completed at a range of torque factors to achieve a range of maximal pedal strokes across a relevant cadence range (i.e., ~80–160 rev∙min⁻¹). The efforts completed on the Lode Excalibur Sport were in the isoinertial mode, which corresponded to resistive torque factors of (0.6, 1.2 and 1.8 Nm∙kg⁻¹), whilst the efforts completed on the SRM ergometer were completed in the isokinetic modes at fixed cadences (80, 120 and 160 rev∙min⁻¹). The order of efforts was randomised. Our previous investigation showed that there was no significant difference in the mechanical variables (i.e., F_opt and P_max) regardless of whether the P-C profiles were constructed from efforts completed in isokinetic or isoinertial ergometer modes (Wackwitz et al., 2020). Finally, cyclists completed the optimised seated 15-s maximal effort at their F_opt, which were identified from the initial set of three seated efforts (see “Data processing”). Cyclists rolled into the 15-s effort at their F_opt at the active recovery workload.

Session 2

Session 2 was comprised of two isokinetic 15-s maximal sprints separated by 25 min of active and passive recovery. The two 15-s sprints were completed at cadences equating to ± 15% of F_opt (i.e., +15%F_opt and −15%F_opt) identified from session 1. Thus, a high cadence effort at +15%F_opt and a low cadence effort at −15%F_opt were completed, the order of which was randomised (Figure 1). Cyclists rolled into the 15-s bouts at the prescribed cadence (±15%F_opt) at the active recovery workload.

Figure 1. Outline of the testing protocol. Bars denote maximal cycling bouts and troughs denote active recovery periods. Panel a denotes session 1 comprised of three 5-s sprints and an optimised 15-s sprint completed at the athletes F_opt. Panel B denotes session 2, comprised of two 15-s sprints at 15% above and below F_opt. Panel C denotes the group P-C relationship and the three cadences in which the 15-s maximal sprints were completed. The apex of the model represents P_max, with the corresponding cadence denoting F_opt. Grey dashed lines represent ±15%F_opt, the cadences prescribed within panel B.

Data processing

To prescribe cadence as ±15%F_opt the P-C profile for each participant had to be constructed. The data processing procedures outlined in (Wackwitz et al., 2020) were utilised to identify the polynomial relationship between power and cadence. Given the importance of accurately identifying P_max and F_opt, third-order polynomial equations were utilised to represent the relationship. F_opt was interpolated as the cadence at which P_max occurred. To quantify the fatigue rate, the maximal data from peak values were plotted and a linear trendline was fitted to represent the data as a function of both time (s) and cumulative pedal strokes. The y-intercept was fixed as the peak value. The slope of the trendline is thus used as the fatigue rate.

Statistical analysis

The de-identified data were analysed using SPSS v26 software (IBM, Armonk, New York, United States) and Prism v9 software (GraphPad, San Diego, California, United States). Descriptive analysis and tests of normality were calculated for all variables. The repeated measures ANOVA was used to compare the relative variables’ relative decrement per second (DecR_sec) and relative decrement per pedal stroke (DecR_str) between the F_opt, +15%F_opt and −15%F_opt trials. The Bonferroni correction for multiple comparisons was applied. Paired t-test analysis was used to compare absolute decrement per second (DecA_sec) and absolute decrement per stroke (DecA_str) between the ±15%F_opt trials. The dataset was inspected for outliers using the ROUT method in Prism (Motulsky & Brown, 2006).

Results

P-C profiles

The P-C profiles established from session 1 yielded an average maximal power output of 1539 (480) W and had a group F_opt of 129 (9) rev∙min⁻¹. The group P-C profile is demonstrated in Figure 1.

Power output during trials

There was a significant difference in the P_peak achieved during the three trials. The F_opt trial had a significantly higher P_peak than the ±15% F_opt trials, whilst the ±15% F_opt trials did not significantly differ. A significant difference in the mean power output (P_mean) was also identified, with the F_opt and −15%F_opt trials having a greater P_mean than the +15%F_opt trial. There was no significant difference in P_mean between the F_opt and −15%F_opt trials (Figure 2).

Figure 2. Panel A denotes the power output during the 15-s maximal sprint cycling bouts at F_opt and ±15%F_opt. Dashed black line represents the linear trendline fitted to the F_opt trial (data represented by open black circles). Black line represents the linear trendline fitted to the −15%F_opt trial (data represented by black circles). Grey line represents the linear trendline fitted to the +15%F_opt trial (data represented by grey circles). Panel B denotes differences in peak and mean power output trials at F_opt and ±15%F_opt. Data presented as mean (standard deviation, SD). Abbreviations: P_peak, peak power output in each trial; P_mean, mean power output over the duration of each trial.

The linear regression analysis identified a significant relationship between P_peak and absolute fatigue rate at −15%F_opt (p = 0.0017, R² = 0.682), F_opt (p < 0.0001, R² = 0.876) and +15%F_opt (p = 0.0008, R² = 0.733). In contrast, no significant relationship was identified between relative fatigue rate and P_peak at −15%F_opt (p = 0.780, R² = 0.009), F_opt (p = 0.395, R² = 0.082) or +15%F_opt (p = 0.879, R² = 0.003). Given the presented evidence that absolute fatigue is confounded by P_max, we recommend that relative fatigue be used in conjunction with absolute rates in future studies to ensure a more comparable and comprehensive metric of fatigue.

Fatigue rate at different cadences

There was a significant difference between the F_opt, +15%F_opt and −15%F_opt trials in the decrement in power output per second. Post-hoc analysis revealed that the decrement in power output per second was significantly lower in the low cadence (−15%F_opt) trial compared to both the F_opt and high cadence (+15%F_opt) trial. However, when the relative decrement was expressed as a function of pedal stroke (i.e., decrement per pedal stroke), there was no significant difference between any of the three trials (Figure 3). Given that efforts were completed at fixed cadences, the relative torque decrement mirrors these findings. The relative decrement in torque per stroke did not significantly differ between the three trials (p > 0.05), while there was a significant difference when expressed as a function of time (Table 1).

Figure 3. Fatigue rates denoting the decrement in normalised torque at different cadences. Panel a displays the decrement as a function of cumulative pedal strokes where the rate of decrement does not significantly differ. Panel B displays the decrement as a function of time, where the rate of decrement significantly differs. Data presented is that of the group average.

Table 1. Differences in the relative decrement between F_opt and ±15%F_opt.

Variable	-15%Fopt (A)	Fopt (B)	+15%Fopt (C)	F-statistic	Sig.	Post hoc
DecRsec	2.2 (0.4)	2.8(0.5)	3.3 (0.5)	28.94	<0.0001	B > A C > A
DecRstr	1.2 (0.2)	1.3 (0.2)	1.3 (0.2)	2.41	0.1162

Data presented as mean (standard deviation, SD). A significant difference between trials is denoted by * (p < 0.05). Abbreviations: DecR_sec, Relative decrement in power output per second; DecR_str, relative decrement in power output per pedal stroke; F_opt, optimal cadence.

As with the relative results, when expressed as an absolute decrement per second there was a significant difference between the +15%F_opt and −15%F_opt trials. However, when the decrement in power output was expressed as an absolute decrement per pedal stroke, there was no significant difference between the +15%F_opt and −15%F_opt trials (Table 2). The paired t-test analysis was completed as the ±15%F_opt trials were matched for P_peak, whereas the F_opt was significantly higher (Figure 2). Given that absolute decrement is related to P_peak, comparing absolute values from the F_opt trials to those of the ±15%F_opt trials is not warranted.

Table 2. Differences in absolute decrement between −15%F_opt and +15%F_opt.

Variable	-15%Fopt	+15%Fopt	Effect size	t-statistic	Sig.
DecAsec	27.9 (9.8)	41.4 (15.5)	1.50	4.974	0.0006*
DecAstr	15.3 (5.1)	16.7 (6.1)	0.41	1.372	0.2002

Data presented as mean (standard deviation, SD). A significant difference between trials is denoted by * (p < .05). Abbreviations: P_peak, peak power output achieved within each trial; DecA_sec, decrement in power output per second (W∙s⁻¹); DecA_str, decrement in power output per pedal stroke (W∙ΔѲ⁻¹); F_opt, optimal cadence.

Discussion

The present study aimed to investigate the effect of cadence on fatigue rate during maximal sprint cycling. Our results indicate that sprint cycling at a cadence 15% higher than optimal (+15%F_opt) produces an absolute and relative decrement in power output per second that is significantly larger than at a cadence 15% lower than optimal (−15%F_opt) despite P_peak being similar between trials. However, when the relative decrement in power output was calculated as a function of cumulative maximal pedal strokes, there was no significant difference between the three trials performed at −15%F_opt, F_opt and +15%F_opt. These findings suggest that the decrement in power output during maximal sprint cycling is primarily determined by the number of cumulative maximal pedal strokes completed. As such, a possible tactic to minimise fatigue during maximal sprint cycling would be to increase the gear ratio, which would result in a decreased cadence. Evidence supporting such a tactic is displayed within our results, whereby the −15%F_opt trial had a significantly lower P_peak than the F_opt trial; however, the average power output achieved over the 15-s trial did not significantly differ. Such a tactic would need to factor both minimising fatigue and maximising the ability to produce power. Additionally, our findings suggest that the fatigue rate is dependent on the number of maximal pedal strokes completed.

The finding that fatigue per pedal stroke is independent of cadence is in accordance with those from Tomas et al. (2010) who reported similar fatigue rates per pedal stroke across different cadences (109–135 rev∙min⁻¹). (Tomas et al., 2010) investigated fatigue rates during 30-s maximal sprint cycling trials at different cadences, with both trials optimised for their respective crank lengths. Their findings suggested that pedalling rate was the key determinant of fatigue rate rather than crank length. However, it was unclear whether the fatigue rate per pedal stroke was equal due to the sprint cycling efforts being completed at the optimal cadence or whether the decrement per pedal stroke was equal. The findings from our study suggest that the relative decrement in power output and torque per pedal stroke, for each individual, is equal. Thus, as shown previously (McCartney et al., 1983; Tomas et al., 2010), lower cadence efforts minimised the fatigue during maximal sprint cycling. Interestingly, the absolute fatigue rate observed in the present study was significantly related to the P_peak in the F_opt and ±15%F_opt trials; however, there was no significant relationship between P_peakand relative fatigue rate.

A decrement in power output being stroke-dependent suggests that one maximal pedal stroke is associated with a similar decrement in power irrelevant of the total work completed. As the two cadence trials (i.e., ±15%) were P_peak matched, power is equal. Given that cadence is a measure of pedal strokes per minute, a higher cadence results in a greater number of cumulative pedal strokes than a lower cadence, within a fixed timeframe. Resultantly, a lower cadence cycling bout will complete less maximal pedal strokes and experience less fatigue. Further research could investigate the fatiguing mechanisms underpinning the stroke-dependent reduction in power output. It should be noted that the findings relating to changes in the P-C relationship are extrapolated from a comparatively narrow, but ecologically valid, cadence range of ± 15%F_opt. This resulted in a cadence range of ~95–160 rev∙min⁻¹. This range was chosen as it is representative of the upper limit a track sprint event would be completed (Dorel et al., 2005). However, it may not provide inference as to how the decrement in torque and power output manifests at more extreme cadences (e.g., towards maximal torque and maximal cadence). As a result, further research investigating the effect of cadence on fatigue rate at cadences further from F_opt is warranted to gain further understanding as to the effect of fatigue on the P-C relationship.

Practical application

The findings presented within this study demonstrate that a lower cycling cadence (within 15% of F_opt) can minimise fatigue during maximal sprint cycling compared to higher cadences. As such, during maximal sprint cycling, cadence and gear selection should not only be optimised to produce high P_peak but also to minimise the number of maximal pedal strokes completed within the sprint duration. Optimal cadence selection during a maximal cycling sprint will likely reflect a balance between maximising P_peak, whilst reducing cadence to minimise the maximal number of pedal strokes completed. Quantifying an individual’s fatigue rate, and P-C relationship as outlined within this investigation, could be used as input variables into a physics-based model of track cycling to calculate an individual athlete’s theoretically optimal gear ratio.

Conclusion

During maximal sprint cycling, the decrement in power output is dependent on the number of cumulative pedal strokes. The decrement is individualised; however, irrelevant of the cadence, the relative decrement in power output each stroke elicited was equal. As a result of this finding, a novel approach to optimising sprint performance could be to select a cadence that both maximises the ability to produce power and minimises the effect of fatigue. Further research could investigate the physiological mechanisms underpinning the cumulative pedal stroke-determined fatigue rate.

Acknowledgments

The authors have no financial relationships relevant to this article to disclose and no competing interests to disclose. The results of the present study are presented clearly, honestly, and without fabrication.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This research was funded by the Queensland Academy of Sport through the Sport Performance Innovation and Knowledge Excellence system.