Field- and Laboratory-derived Power-Cadence Profiles in World-Class and Elite Track Sprint Cyclists

ABSTRACT

Previous investigations comparing Torque-Cadence (T-C) and Power-Cadence (P-C) profiles derived from seated and standing positions and field and laboratory conditions are not congruent with current methodological recommendations. Consequently, the aim of this investigation was to compare seated and standing T-C and P-C profiles generated from field and laboratory testing. Thirteen world-class and elite track sprint cyclists (n = 7 males, maximal power output (P_max) = 2112 ± 395 W; n = 6 females, P_max = 1223 ± 102 W) completed two testing sessions in which field- and laboratory-derived T-C and P-C profiles were identified. Standing P-C profiles had significantly (p < 0.05) greater P_max than seated profiles, however there were no significant differences in optimal cadence (F_opt) between seated and standing positions. P_max and F_opt were significantly lower in field-derived profiles in both positions compared to laboratory-derived profiles. However, there was no significant difference in the goodness-of-fit (R²) of the P-C profiles between laboratory (0.985 ± 0.02) and field-testing (0.982 ± 0.02) in each position. Valid T-C and P-C profiles can be constructed from field and laboratory protocols; however, the mechanical variables derived from the seated and standing and field and laboratory profiles cannot be used interchangeably. Both field and laboratory-derived profiles provide meaningful information and provide complementary insights into cyclists’ capacity to produce power output.

KEYWORDS:

• Elite
• sprint cycling
• power profiles
• field-testing

Introduction

Given the importance of maximal power output (P_max) in elite track sprint cycling (Dorel et al., 2005), Torque- and Power-Cadence (T-C, P-C) profiling has been used to estimate the theoretical value of the highest power output that can be produced by a cyclist with optimal mechanical constraints (i.e., optimal cadence (F_opt) and optimal torque factor) (Dorel et al., 2005; Gardner et al., 2005; Kordi et al., 2019; Rudsits et al., 2018; Wackwitz et al., 2020). The quantification of T-C and P-C profiles allows for the identification of these mechanical variables which are important for talent identification, monitoring athlete progression, and optimising gear selection in track-cycling events (Coyle et al., 1981; Dorel et al., 2005; Coyle et al., 1979; Wackwitz et al., 2020). The methodological considerations for deriving T-C and P-C profiles are now well established with recommendations available as to the number of sprints, as well as data processing and modelling procedures to construct these profiles (Dorel, 2018; Rudsits et al., 2018; Wackwitz et al., 2020). However, limited research has demonstrated the compatibility of laboratory-based T-C and P-C profiling with field derived profiling (Bertucci et al., 2005; Gardner et al., 2007). Establishing the level of agreement between laboratory- and field-derived profiles may improve the ecological validity of field-based T-C and P-C profiles and provide additional insight into the relevance of these profiles to real-world performance scenarios (Dorel, 2018).

Construction of valid T-C and P-C profiles is necessary to estimate P_max (Dunst et al., 2022) and recommended methodological considerations, including a minimum of three short duration efforts (Dorel, 2018), data processing procedures including the filtering of non-maximal and fatigue data (Rudsits et al., 2018; Wackwitz et al., 2020), and the utilisation of third-order polynomial regression to represent the maximal data, have been proposed (Rudsits et al., 2018; Wackwitz et al., 2020). Recently, the importance of the range of cadences used to represent the relationship has been demonstrated (Dunst et al., 2022; Wackwitz et al., 2020). While it is well accepted that a polynomial relationship exists between power and cadence (Dorel et al., 2005; Gardner et al., 2007; Kordi et al., 2019; Rudsits et al., 2018; Wackwitz et al., 2020; Yeo et al., 2015), there is contention as to which order polynomial equation best represents the P-C relationship. Despite second-order polynomials being used extensively within the literature (Dorel et al., 2005; Kordi et al., 2019), recent studies have shown that third-order polynomial models provide significantly better goodness-of-fit representing the P-C relationship (Arsac et al., 1996; Rudsits et al., 2018; Wackwitz et al., 2020; Yeo et al., 2015). With this contemporary set of recommendations, many studies are no longer in accordance. As a result, the accuracy of previously quantified field-based T-C and P-C profiles is unknown, warranting further investigations comparing field- and laboratory-based protocols.

Despite many studies investigating T-C and P-C profiles in laboratory settings (Dorel et al., 2005; Kordi et al., 2019; McCartney et al., 1983; Rudsits et al., 2018; Wackwitz et al., 2020; Yeo et al., 2015), few have attempted to evaluate field-based approaches to determine T-C and P-C profiles (i.e., on the track) (Bertucci et al., 2007; Dunst et al., 2022; Dwyer et al., 2022; Gardner et al., 2007). Gardner et al. (2007) reported that there were no significant differences in the T-C and P-C profile mechanical variables derived from a laboratory- and field-based protocol in elite cyclists. This study utilised a protocol incorporating three short-duration laboratory sprints and two standing start field tests. These findings may be somewhat surprising given the lower cadence range of the stationary-start field testing sprints (50–100 rev⋅min⁻¹) compared to the laboratory sprints (50–125 rev⋅min⁻¹). Consequently, rolling or motor-assisted efforts could be employed to increase the range of cadences achieved in the field-testing sessions, while also minimising within-sprint fatigue. In addition, the data processing procedures utilised within their investigation are not in accordance with current recommendations; in regard to the data filtering process, number of sprints completed and the polynomial representation of the P-C relationship (Dorel, 2018; Rudsits et al., 2018; Wackwitz et al., 2020). While Gardner et al. (2007) provided valuable insight into the feasibility of field testing, further research is required to investigate whether the mechanical variables of the T-C and P-C profiles derived from standing and seated as well as field and laboratory are comparable while employing current best-practice methodologies.

Bertucci et al. (2005) investigated power output and force production during field and laboratory testing in both the seated and standing positions, however, did not construct P-C profiles, and thus values of P_max and F_opt were not identified. Similarly, the high-frequency torque and cadence data were averaged over 0.1-s intervals, which increases the variability of the force data, given that different sections of the pedal stroke have vastly different torque production (Bini et al., 2013; Hansen et al., 2012). Therefore, the methodology employed is not in accordance with current recommendations suggesting that data should be stroke averaged (Dorel, 2018). Nonetheless, Bertucci et al. (2005) did establish that peak power output in the seated position was higher in laboratory testing and in the standing position was higher in field testing. Yet, without constructing valid T-C and P-C profiles it is unclear whether the cadences achieved between the two settings were equivalent or whether P_max was comparable. As such, we believe that employing an evidence-based methodology with best-practice data processing procedures to generate T-C and P-C profiles in both the field and laboratory as well as during seated and standing positions will provide more clarity as to whether valid field-based T-C and P-C profiles can be established. Therefore, the aim of the present study is to compare field- and laboratory derived T-C and P-C profiles in both the seated and standing positions. We hypothesise that valid field-derived T-C and P-C profiles could be constructed with a testing protocol concurrent with methodological recommendations, and that P_max and F_opt would not significantly differ from laboratory-derived profiles.

Materials and methods

Participants

Six female (20.0 ± 2 yr, 70.3 ± 5.53 kg) and seven male (22.0 ± 3.5 yr, 87.8 ± 7.35 kg) track-sprint cyclists, free from known injury or illness, volunteered for the present study. Personal best times for the flying 200 m (f200 m) track-cycling event for female cyclists (n = 6) were between 10.64 to 11.41 s, 9.53 and 9.975 s for male cyclists (n = 6), and 10.78 s for the male para-cyclist. All cyclists had competed at international-level track cycling events. Three cyclists were classified as world-class athletes with podium places at major international track-sprint events, with the remaining ten cyclists being classified as elite (McKay et al., 2022). Prior to participation all cyclists provided written informed consent according to the Declaration of Helsinki and the study was approved by the Griffith University Human Research Ethics Committee.

Participant recruitment

This investigation utilised purposive sampling, given the strict inclusion criteria and low target population. To meet the inclusion criteria for this investigation participants must at least: 1) have competed at an international level track cycling competition, and 2) currently be completing maximal training, within the given sports norms, with intention to complete at top-level competition. These criteria align with the classification framework proposed by Mckay et al. (2022). Such participants would be within the top 300 in the world for their relevant events.

Design

This project utilised a within-subject design in which cyclists attended one laboratory-testing session at the Queensland Academy of Sport laboratory (Nathan, Queensland, Australia) or Adelaide Superdrome (Gepps Cross, South Australia, Australia), and one field-testing session at the Anna Meares Velodrome or Adelaide Superdrome separated by a minimum of 2 days and up to 7 days. However, one cyclist, due to COVID-related interruptions, was unable to complete “Session 2” within the required timeline and as such their data were excluded from the field- and laboratory-testing comparisons.

Methodology

Cyclists who completed their laboratory testing session at the Queensland Academy of Sport laboratory performed their trial on an electrodynamically braked cycle ergometer with force pedals (Lode Excalibur Sport, Lode B. V., Groningen, the Netherlands) with a data output of one data point for every two degrees of each pedal revolution, while the Adelaide Superdrome housed a custom motor-driven SRM ergometer (SRM, Jülich, Germany) with an output of 200 Hz. The ergometer was set up for each cyclist based on their competition bicycle setup. A standard crank length of 170 mm was set, and cyclists used their own cleated shoes while the pedals were kept constant. Cyclists who completed their field-testing session at the Anna Meares Velodrome utilised a Verve InfoCrank (Track InfoCrank 144BCD ISIS, VerveCycling, Berkshire, United Kingdom) powermeter with the datalogger capable of extracting data at 256 Hz, while field-testing session at the Adelaide Superdrome utilised an SRM power metre (PM9, SRM, Jülich, Germany) capable of 200 Hz.

Laboratory session

Cyclists completed a warmup protocol that comprised cycling at 50, 100 and 150 W for 120 s each. This was followed by two, 6-s maximal effort cycling sprints, separated by a 240-s recovery period where participants cycled at 100 W and 150 W for 120 s each (Wackwitz et al., 2020). Prior to the start of the experimental test (i.e., the “power profile”), cyclists were asked to actively recover for 5 min at a self-selected power output.

The laboratory session consisted of six, 5-s maximal sprints (three in seated and standing position), an additional 5-s maximal standing sprint, and a maximal 15-s seated sprint performed at the F_opt (see Figure 1). The three seated and three standing 5-s maximal sprints were completed at 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, 160 rev∙min⁻¹). The order of the efforts was randomised and paired between positions. 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 T-C and P-C profiles were constructed from efforts completed in isokinetic or isoinertial ergometer modes (Wackwitz et al., 2020). Finally, cyclists completed two optimised sprints; a standing 5-s and a seated 15-s maximal effort which were completed at the athletes’ F_opt, which were identified from the initial set of three standing or seated efforts (see “Data processing”). The identified F_opt was prescribed as the cadence for both the 5-s standing and 15-s seated optimised efforts in isokinetic mode.

Figure 1. Laboratory session schematic. Data utilised to construct laboratory-derived, seated and standing T-C and P-C profiles.

Field session

The field session consisted of six, 65-m maximal sprints, three in seated and three in standing position (Figure 2). The three seated and three standing 65-m sprints were completed on the velodrome separated by 6 min of passive and active recovery. To achieve a range of cadences comparable to laboratory testing within these sprints, each sprint was completed at a different roll-in speed; a walking pace roll-in, a motor-paced lead-in at 30 km⋅h⁻¹ and a motor-paced lead-in at 50 km⋅h⁻¹. Each effort had a targeted cadence of 80, 120, and 160 rev∙min⁻¹, respectively. Data processing procedures are detailed in the “Data processing” section.

Figure 2. Field-testing session schematic. Data utilised to construct field-derived, seated and standing T-C and P-C profiles.

Data processing

Data from the ergometers and power metres were stroke averaged using a custom-written Python script (v3.8). Stroke-averaged data was used to construct fatigue-free T-C and P-C profiles for both the seated and standing position emanating from Session 1 and 2, following the methodology outlined in our previous investigation (Wackwitz et al., 2020). Polynomial regression analysis using a third-order polynomial equation with the y-intercept constrained to equal zero was utilised to represent the maximal filtered data (Rudsits et al., 2018; Wackwitz et al., 2020). The polynomial equation was then solved to identify P_max and interpolate the corresponding cadence (F_opt).

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 dataset was inspected for outliers using the ROUT method in Prism (Motulsky & Brown, 2006). A repeated measures two-way analysis of variance (ANOVA) was used to compare P_max and F_opt values for the seated and standing positions obtained from field and laboratory testing. Tukey’s post-hoc test was applied to identify differences when the two-way ANOVA indicated a significant interaction effect. The level of significance was set at p < 0.05. Effect size for paired samples were calculated to detail the magnitude of effect between field and laboratory derived profiles in seated and standing positions (Cohen, 1988; Dankel & Loenneke, 2021). The interpretation of effect sizes are 0.2: small, 0.6: moderate, 1.2: large, 2: very large, 4: extremely large (Hopkins et al., 2009). Goodness of fit (R²) was calculated for the third order polynomial regression trendlines. Paired t-test analysis was used to compare the goodness of fit for laboratory- and field-derived P-C profiles. Effect size for paired samples were also calculated for the difference in goodness of fit of the polynomial regression equations between field and laboratory derived P-C profiles.

Results

P_max values established were significantly higher in laboratory than field testing (Fx,y = 10.81, P = 0.0072) and significantly higher in the standing position in comparison to the seated position (Fx,y = 60.54, P < 0.0001). However, no significant interaction effect on P_max between setting and position was found (Fx,y = 4.833, P < 0.0502). Similarly, the F_opt values were significantly higher in laboratory testing compared to field testing (Fx,y = 17.21, P = 0.0016), however, there was no significant difference in F_opt between the seated and standing position (Fx,y = 1.888, P = 0.198). Additionally, no significant interaction effect on F_opt between setting and position was found (Fx,y = 0.2394, P < 0.6350) (See Tables 1 and Figure 3).

Figure 3. Group laboratory- (left) and field-derived (right) P-C profiles, in seated and standing positions.

Table 1. Tukey post-hoc analysis investigating the effects of testing location (laboratory, field) and cycling position (seated, standing) on P_max and F_opt. Data presented as mean (standard deviation, SD). Effect size = ES. * denotes P < 0.05, ** denotes P < 0.01, *** denotes P < 0.001.

	Laboratory		Field
Variable	Seated (A)	Standing (B)	Seated (C)	Standing (D)	Post-hoc
Pmax	1556 (460)	1758 (530)	1361 (367)	1622(455)	A < B*** (ES = 2.35), A > C*** (ES = 1.14) B > D*** (ES = 0.72), C < D*** (ES = 2.02)
Fopt	129 (9)	125 (10)	120.4 (13)	119 (10)	A > C** (ES = 0.90) B > D** (ES = 1.21)

Upon completion of the outlined statistical analysis, further investigation into individual differences in field and laboratory P-C profiles was warranted. As such, a ranking analysis was completed to investigate whether there was an interaction between participant and the results emanating from field and laboratory testing (Figures 4,5). The drop in P_max from laboratory testing to field testing is not consistent within all participants. Results presented within Figures 4 and 5 indicate that specific participants are able to produce similar P_max in field and laboratory-testing session, however other participants are not able to.

Figure 4. Individual differences between field and laboratory-derived P_max values for seated (left panel) and standing (right panel) positions.

Figure 5. Cyclist’s ranking for field and laboratory-derived P_max values during seated (left panel) and standing (right panel) efforts. Ranking one indicates the cyclist with the highest P_max.

Model Validity

There was no significant difference in goodness-of-fit parameters between the laboratory- and field-derived seated P-C profiles. Similarly, there was no significant difference in the goodness-of-fit between laboratory- and field-derived standing P-C profiles (Table 2).

Table 2. Differences in field vs laboratory testing model goodness-of-fit (n = 13). Data presented as mean (standard deviation, SD).

Variable	Laboratory (R^2)	Field (R^2)	Effect size	t-statistic	Sig.
Seated	0.9865 (0.015)	0.9849 (0.018)	0.07	0.2636	0.7966
Standing	0.9841 (0.025)	0.9795 (0.019)	0.137	0.4945	0.6299

Discussion

Our findings indicate that T-C and P-C profiles constructed from laboratory testing had significantly higher F_opt and P_max than the field-derived profiles in both seated and standing positions. Additionally, we found that P_max differs between seated and standing cycling positions in both the laboratory and the field. F_opt established from field and laboratory testing did not significantly differ between the seated and standing positions. Despite the differences in mechanical parameters derived from these models, both field-derived and laboratory-derived P-C profiles had high R² values that did not significantly differ and had a small magnitude of difference. As such, we believe valid P-C profiles can be constructed from both field and laboratory testing; however, high ability to produce power output in a laboratory setting does not necessarily translate into a field setting.

Our findings show that P_max and F_opt values derived from laboratory testing are significantly higher than those derived from field testing. We postulate that the differences in P_max and F_opt between field and laboratory-derived P-C profiles are due to the increased complexity of riding on the track and the altered biomechanics affecting the application of force, especially within the standing position. The ergometers employed within this investigation do not permit lateral movement or sway, as is often adopted when cyclists utilise the standing position on the track (Dorel, 2018; Faria et al., 2005b). The lateral sway observed during standing sprint cycling alters both muscle activation patterns (Faria et al., 2005a, 2005b), the application of force and additional muscle(s) groups are recruited (Li & Caldwell, 1998; So et al., 2005). Nonetheless, the application of force through the pedals to produce power and propel the cyclist forward may be diminished as the application of force will not align with the direction of the pedal stroke. In addition to the altered muscle activation patterns and biomechanics, cycling on a velodrome is a more complex task requiring enhanced balance and motor coordination, and the skill set required for cornering and cycling around a velodrome, as well as the perceived risk associated with cycling at speeds exceeding 70 km·h⁻¹ are likely to influence task execution (Pijpers et al., 2005).

The finding that mechanical parameters interpolated from field- and laboratory-derived P-C profiles significantly differ, contrasts with previous findings (Gardner et al., 2007). The findings from Gardner et al. (2007) illustrate no significant differences between field- and laboratory-derived measures of maximum torque (T_max), P_max or F_opt. The methodology utilised by Gardner et al. (2007) is not concurrent with current recommendations (Dorel, 2018; Rudsits et al., 2018). Nonetheless, the linearity in the T-C profiles suggests that the data was not affected by non-maximal (i.e., fatigued) data points (Gardner et al., 2007). However, the modelling procedures used to represent the P-C relationship are no longer best practice as recent recommendations suggest that third-order polynomial equations may represent the maximal P-C profile more accurately (Rudsits et al., 2018; Wackwitz et al., 2020). This may limit the precision of the estimated P_max and F_opt values in Gardner et al. (2007). Additionally, given the low peak cadence achieved within field testing, it seems reasonable that T_max identified from field testing within Gardner et al. (2007) accurately represents the athlete’s maximal ability to produce torque, however, due to extrapolating P_max and F_opt from well outside the cadence range achieved within testing, these values may have less precision.

The findings emanating from this project are in agreement with those presented by Dwyer et al. (2022) that P_max and F_opt significantly differ between seated and standing positions. However, part of the disparity between their P-C profiles is likely caused by the seated profiles being completed in a laboratory setting whilst the standing profiles were derived from field testing. As such, care should be taken when comparing mechanical parameters such as P_max and F_opt between T-C and P-C profiles constructed from field and laboratory testing. In accordance with previous investigations (Dwyer et al., 2022), our findings demonstrate that P_max is significantly higher in the standing position compared to seated. This relationship was evident in both field- and laboratory-derived P-C profiles. This is consistent with findings that peak power output is greater in standing position in road cyclists (Merkes et al., 2020). It has been postulated that this is likely due to differences in the contribution of muscular and non-muscular forces (Dwyer et al., 2022; Kautz & Hull, 1993), altered muscle recruitment patterns, and joint kinematics (Wilkinson et al., 2020).

Our findings demonstrate that during laboratory testing, cyclists were able to produce greater P_max in both the seated and standing positions compared to field testing. Interestingly, when their position was ranked within the participant pool and compared between field and laboratory testing there were substantial changes in the rank order (Figure 4 & 5). This suggests that some elite track sprint cyclists are able to maintain their ability to produce P_max on the track better than others. The changes in ranking were greater in the standing position compared to seated. Given there is a larger technical requirement for standing sprinting, we suggest the difference between field and laboratory-derived P_max values may have a technical basis. As such, further research could investigate the deficit between laboratory- and field-derived P_max and whether it is related to training age or the effect of targeted technical training. While our investigation provides valuable insight into the differences between field and laboratory-derived T-C and P-C profiles, there are potential limitations that must be considered when interpreting the findings. The potential difference in equipment used through field and laboratory testing could affect the evidence presented in this investigation. However, both the SRM ergometer and LODE Excalibur sport have exceptional precision and have been used as gold standard measures of torque, power, and cadence (Kordi et al., 2019; Reiser et al., 2000; Rudsits et al., 2018; Wackwitz et al., 2020). Similarly the SRM and InfoCrank powermeters utilised for field testing have been validated and used as a gold standard benchmark for validating other powermeters (Bertucci et al., 2005; Merkes et al., 2019).

Practical applications

This investigation provided insight into cyclist’s ability to produce force and power maximally in field and laboratory settings. We propose the protocols and metrics presented within this paper can be utilised to monitor changes in performance, optimise performance, and for talent identification (Coyle et al., 1981; Dorel et al., 2005; Wackwitz et al., 2020). Laboratory testing removes the technical component of cycling on the velodrome and thus provides data that could be considered relevant for the field-based performance potential be efficacious for talent identification and tracking responses to training interventions. Field-based measures may be more relevant to competition-specific performance and should be used for informing tactical competition decisions including gear selection and performance modelling. Integrating both laboratory- and field-derived profiling methods may highlight the need for technical improvements to maximise the transferability of laboratory-derived testing into the field.

Conclusion

Despite mechanical parameters associated with P-C profiles significantly differing between field and laboratory testing, we propose that both profiles emanating from field and laboratory testing are valid. Field-derived profiles are likely more related to event performance; however, laboratory testing can remove the technical element required for field-derived profiling and offer insight into field-based performance potential. We propose that both field and laboratory-derived profiles provide meaningful information and provide complementary insights into cyclists’ capacity to produce power output.

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 authors.

Additional information

Funding

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