A self-paced 15-minute cycling time trial is a reliable performance measure in recreationally active individuals

ABSTRACT

Cycling time trial (TT) protocols have been shown to be reliable in trained cyclists, but their reproducibility in lesser-trained individuals is unknown. This study examined the reliability of a self-paced 15-minute cycling TT in recreationally active individuals. Twelve recreationally active males (age 27 ± 3 y; body mass 75.2 ± 8.9 kg; $\dot{\mathrm{V}}$O_2peak = 51.10 ± 7.53 ml∙kg∙min⁻¹) completed a $\dot{\mathrm{V}}$O_2peak test and four experimental trials, separated by > 48 h. Experimental trials consisted of 10 min cycling at 60% W_max, followed by a self-paced 15-min TT. Heart rate and work done were recorded every 5 min during the TT; and coefficient of variation (CV) was calculated. Work done was not different (P = 0.706) between trials (193.2 ± 45.3 kJ; 193.2 ± 43.5 kJ; 192.0 ± 42.3 kJ; 193.9 ± 42.8 kJ). Within participant CV ranged from 0.5–4.9% for the four TTs, with a mean CV of 2.1%. Mean CV decreased from 2.0% (range 0.1–5.0%) for the first two TTs to 1.7% (range 0.2–5.6%) for the second and third TTs, and further decreased to 1.0% (range 0.2–1.8%) for the third and fourth TTs. In conclusion, the use of a short-duration self-paced cycling TT in recreationally active individuals is a reliable performance measure.

KEYWORDS:

• Exercise testing
• performance
• coefficient of variation
• training status
• reproducibility
• validity

Introduction

Performance tests are commonly used to evaluate training, nutritional, and other experimental interventions, within sport and exercise science research. Ensuring these laboratory-based performance tests have high reproducibility is essential to determine whether differences in exercise performance are attributable to an experimental intervention, rather than measurement error or intraindividual variation (Atkinson & Nevill, 2001). Traditional performance testing methods include time-to-exhaustion (TTE) and time trial (TT) protocols. Time-to-exhaustion protocols consist of participants exercising at fixed sub-maximal exercise intensities (i.e., a percentage of $\dot{\mathrm{V}}$O_2peak or W_max) until they can no longer sustain the necessary power output or speed. In contrast, time trial protocols have a known endpoint, with participants required to either complete as much work as possible within a given time period, or complete a set distance/amount of work as quickly as possible (Laursen et al., 2007).

Time-to-exhaustion protocols are commonly used to investigate physiological mechanisms during exercise (Currell & Jeukendrup, 2008), however, some studies report large within-participant variability (Jeukendrup et al., 1996; McLellan et al., 1995). Conversely, time trial protocols simulate performance more accurately (Palmer et al., 1996), and have been shown to be reproducible (demonstrated by a coefficient of variation of 1–4%) in trained runners (Laursen et al., 2007; Rollo et al., 2008; Schabort et al., 1998) and cyclists (Bishop, 1997; Hickey et al., 1992; Jeukendrup et al., 1996; Laursen et al., 2003; MacInnis et al., 2019; Palmer et al., 1996; Smith et al., 2001; Sparks et al., 2016; Sporer & McKenzie, 2007).

Despite the large volume of research confirming the reliability of cycling time trials in trained cyclists, the reliability of such protocols in lesser-trained, recreationally active individuals is unknown. As recreationally active individuals are commonly recruited for sport and exercise science research (Cheuvront et al., 2009; Clayton et al., 2015; Kenefick et al., 2009), determining the reliability of time trial protocols in this population group is of great importance. Additionally, the accurate assessment of performance is necessary in strength and conditioning/gymnasium and laboratory-based settings, to determine the effectiveness of interventions, whether it be a training intervention for an active client/athlete, or an experimental intervention addressing a research question. A short-duration self-paced time trial is a relatively simple performance test to implement in an applied sports environment, and therefore, could be used to determine the effectiveness of training interventions and/or monitor a clients/athletes progression throughout a training period. Therefore, the purpose of the present study was to investigate the reliability of a self-paced 15-minute cycling time trial in recreationally active individuals.

Materials and methods

Participants

Twelve recreationally active males (age 27 ± 3 y, body mass 75.2 ± 8.9 kg, height 1.78 ± 0.07 m, BMI 23.7 ± 1.7 kg∙m⁻², $\dot{\mathrm{V}}$O_2peak = 51.10 ± 7.53 ml∙kg⁻¹∙min⁻¹, W_max = 272 ± 50 W) volunteered for this study, which was approved by the Loughborough University Ethics Approvals (Human Participants) Sub Committee (reference number: R15-P088). Using an estimated ICC value of 0.95, statistical power of 80%, significance level of 0.05, 4 trials/repeats, and a null ICC value of 0.8, it was estimated that a sample size of 11 participants would be required (Borg et al., 2022). Prior to commencement of the study, participants provided written informed consent and completed a medical screening questionnaire. Participants were non-smokers, and undertook regular physical activity (at least three sessions of moderate exercise >30 minutes per week). Participants were not trained cyclists and had no history of competing in cycling competitions. Each participant completed a preliminary trial and four experimental trials, separated by a minimum of 48 h.

Pre-trial standardisation

To ensure similar metabolic conditions prior to each experimental trial, all participants recorded their dietary intake and habitual physical activity for the 24 h preceding their first experimental trial. These diet and activity patterns were replicated prior to all subsequent trials and adherence to these requirements was verbally checked. Participants also refrained from any strenuous exercise or alcohol intake during this period. The specific time of trials was standardised for each participant and kept constant across all trials. Participants were instructed not to consume any food or fluid in the 2 h prior to arrival at the laboratory, with the exception of 500 mL of water 1.5 h before arrival.

Preliminary trial

Prior to the experimental trials, participants completed an incremental maximal exercise test to volitional exhaustion on a cycle ergometer (Lode Corival, Groningen, The Netherlands) to determine peak oxygen uptake ($\dot{\mathrm{V}}$O_2peak) and maximum power output (W_max). Participants initially cycled at 95 W for 3 min, followed by 35 W increments every 3 min until volitional exhaustion. W_max was calculated using the following formula:

W_max = W_com + (t/180)∙35)

Where W_com is the workload of the last completed stage and t is the time in seconds in the final uncompleted stage (Kuipers et al., 1985). These data were used to establish the workload required to elicit 60% and 90% W_max, which were later employed during the experimental trials. Participants determined the saddle height and position, handlebar height and reach, which were recorded and replicated during all subsequent trials. After completion of the maximal exercise test, participants rested for ~20 min before familiarising themselves with the cycle ergometer controls. The controls to adjust the workload on the cycle ergometer were explained to the participants; they started exercise at 90% W_max and cycled for 7.5 min to practice adjusting the exercise intensity. The cycle ergometer monitor was not covered during this period, allowing participants to comprehend the changes in workload.

Experimental trials

Upon arrival at the laboratory, participants voided the contents of their bladder into a plastic container, and nude body mass was recorded (Adam Equipment Co., AFW-120K, UK). This urine sample was analysed for osmolality (Osmocheck, Vitech Scientific, UK) to assess hydration status before trials. A urine osmolality of <900 mOsm∙kgH₂O⁻¹ was taken to indicate the absence of hypohydration (Armstrong et al., 2010) and no participant produced a urine sample greater than this cut off. Thereafter, participants entered a climatic chamber (20.0 ± 0.3°C; 54.2 ± 2.0% relative humidity), mounted a cycle ergometer, and completed a 10 min preload at 60% W_max (163 ± 30 W) followed by a self-paced 15 min time trial (TT). Participants self-selected their cadence for both the preload and TT. Facing air flow (2.4 ± 0.3 m∙s⁻¹) was provided by an electronic fan. Heart rate (Polar FS1, Polar, Finland) and rating of perceived exertion (RPE) were recorded in the final minute of the preload (Borg, 1982). There was a 2 min period between the preload and TT to programme the cycle ergometer and provide standardised instructions. Participants began the TT at a workload corresponding to 90% W_max (245 ± 45 W), but were instructed to complete as much work as possible and could alter the workload (W) freely throughout the TT. The only information available to the participant was the time elapsed, with all other information (work done, heart rate and cadence) blinded from the participant. No verbal encouragement was provided during the TT, and a screen was placed between the investigator and the participant to minimise any peripheral distractions. All trials took place on the same cycle ergometer to minimise any potential systematic error (Paton & Hopkins, 2001). Work done (kJ) and heart rate were recorded every 5 min during the TT.

Statistical analysis

Data were analysed using IBM SPSS Statistics v.22 (SPSS Inc, Chicago, IL) and were initially checked for normality of distribution using a Shapiro-Wilk test. Heart rate, RPE and performance data were analysed using repeated measures analysis of variance (ANOVA). Where the assumption of sphericity was violated, the degrees of freedom were corrected using the Greenhouse-Geisser estimate. Significant results were followed-up by post-hoc paired t-tests or Wilcoxon Signed Rank tests, as appropriate, and the familywise error rate was controlled using the Holm-Bonferroni adjustment. Data sets were accepted as being significantly different when P ≤0.05. Relative test-retest reliability was assessed for total work done with the interclass correlation coefficient (ICC) using a two-way mixed effects model, and was interpreted as: questionable, 0.7–0.8; good, 0.8–0.9; and excellent, >0.9 (Vincent, 1994). Absolute test-retest reliability for work done was determined by calculating the coefficient of variation (CV). The CV between TTs was calculated as ((SD/mean)∙100) for each participant and was averaged to obtain an overall CV. Cohen’s d effect sizes for work done, CVs and pacing during the TTs were calculated as ((Mean 1 – Mean 2)/pooled SD), and were interpreted as: trivial, <0.2; small, 0.2–0.5; medium, 0.5–0.8; and large, >0.8 (Cohen, 1992). All data are represented as mean ± SD, as well as range for some variables.

Results

Pre-trial measures

Pre-trial body mass (P=0.379) and urine osmolality (P=0.867) were not different between trials (Table 1), indicating a similar hydration status at the beginning of trials.

Table 1. Pre-trial body mass and urine osmolality, and heart rate and rating of perceived exertion during the preload for the four trials.

	Trial
	1	2	3	4
Pre-Trial
Body Mass (kg)	74.8 ± 8.9	74.5 ± 9.1	74.5 ± 9.0	74.5 ± 9.0
Urine Osmolality (mOsm∙kgH2O^−1)	355 ± 246	311 ± 196	354 ± 254	326 ± 284
End of Preload
Heart Rate (beat∙min^−1)	143 ± 13	144 ± 15	143 ± 17	140 ± 15
RPE (6–20)	12 ± 1	12 ± 1	12 ± 1	12 ± 1

Data are mean ± SD. RPE = rating of perceived exertion. P > 0.05 for all comparisons between trials.

Performance data

There was no difference between trials for ambient temperature (P = 0.304), relative humidity (P = 0.360) or facing wind speed (P = 0.441).

Work done was similar (P = 0.706; d = 0.00–0.05 [trivial]) between the four TTs (Table 2), and displayed excellent relative test-retest reliability (ICC = 0.99; 95% CI 0.97–1). Within participant CV ranged from 0.5% to 4.9% for the four TTs, which resulted in a mean CV of 2.1%.

Table 2. Work done (kJ) by individual participants for the four 15-min time trials.

	Trial (kJ)						CV (%)
Participant	1	2	3	4	Mean	SD	TT1-TT2	TT2-TT3	TT3-TT4
1	225.3	227.2	226.4	230.7	227.4	2.3	0.6	0.2	1.3
2	249.8	254.2	255.9	255.3	253.8	2.8	1.2	0.5	0.2
3	253.0	239.5	226.3	231.7	237.6	11.6	3.9	4.0	1.7
4	152.3	158.1	155.8	157.3	155.9	2.6	2.6	1.0	0.7
5	265.8	264.5	265.2	267.7	265.8	1.4	0.3	0.2	0.7
6	175.1	168.6	170.0	171.0	171.2	2.8	2.7	0.6	0.4
7	183.9	184.1	181.3	182.6	183.0	1.3	0.1	1.1	0.5
8	162.6	161.8	161.1	159.2	161.2	1.5	0.3	0.3	0.8
9	170.5	183.1	169.2	167.6	172.6	7.1	5.0	5.6	0.7
10	140.0	136.2	141.8	145.1	140.8	3.7	1.9	2.8	1.6
11	137.6	143.7	149.0	152.1	145.6	6.4	3.1	2.6	1.5
12	202.1	197.3	201.6	206.8	202.0	3.9	1.7	1.5	1.8
Mean	193. 2	193. 2	192.0	193.9	193.1	3.9	2.0	1.7	1.0
SD	45.3	43.5	42.3	42.8	-	-	-	-	-

Data are mean ± SD. TT1-TT2 = CV between first and second time trials; TT2-TT3 = CV between second and third time trials. TT3-TT4 = CV between third and fourth time trials.

To examine the impact of familiarisation trials, the CV of pairs of TTs were examined with no familiarisation trials (i.e., the CV between the first [TT1] and second [TT2] time trials), one familiarisation trial (i.e., the CV between the second [TT2] and third [TT3] time trials) and two familiarisation trials (i.e., the CV between the third [TT3] and fourth [TT4] time trials). With no familiarisation (i.e., TT1 vs TT2), the mean CV was 2.0%, with a range of 0.1–5.0%. The mean CV after one familiarisation (i.e., TT2 vs TT3) was 1.7%, with a range of 0.2–5.6%, and was not significantly different (P = 0.450; d = 0.16 [trivial]) to no familiarisation. The mean CV tended to decrease after two familiarisation trials (i.e., TT3 vs TT4) to 1.0%, with a range 0.2–1.8%, (P = 0.051; d = 0.84 [large]) compared to no familiarisation. Compared to one familiarisation (i.e., TT2 vs TT3), the mean CV non-significantly decreased (P = 0.195; d = 0.56 [medium]) after two familiarisations (i.e., TT3 vs TT4). All participants had a CV of less than 2% after two familiarisation trials.

            

In all trials, participants completed more work (P≤ 0.035; d = 0.26–0.41 [small]) at the beginning of the TT (0–5 min: 67.4 ± 12.9 kJ) compared to the middle (5–10 min: 61.8 ± 14.6 kJ) and end of the TT (10–15 min: 63.8 ± 15.1 kJ; Figure 1). There was a tendency (P = 0.086; d = 0.14 [trivial]) for participants to complete more work at the end of the TT (10–15 min) compared to the middle segment (5–10 min). There was a time by trial interaction effect (P = 0.009) for pacing, however, post-hoc tests revealed no differences (P > 0.118).

Figure 1. Work done (kJ) for the three 5-min segments during the four time trials. # indicates significantly (P < 0.05) different from 5–10 min and 10–15 min.

Heart rate and RPE

There were no differences between trials (P > 0.167) for heart rate or RPE during the 10 min preload (Table 1).

During all four TTs, there were no differences (P > 0.999) in heart rate between 5 and 10 min, however, heart rate increased (P < 0.001) at 15 min (Figure 2). There was a time by trial interaction effect (P = 0.015) for heart rate, with post-hoc tests revealing a difference (P = 0.040) at 5 min between trial 1 (171 ± 13 beat∙min⁻¹) and trial 3 (166 ± 12 beat∙min⁻¹), however, no other differences were found (P > 0.098).

Figure 2. Heart rate (beat∙min⁻¹) during the four time trials. # indicates significantly different (P < 0.05) from 5 min and 10 min.

Discussion

There is a large volume of research confirming the reliability of cycling time trials in trained cyclists (Bishop, 1997; Hickey et al., 1992; Jeukendrup et al., 1996; Laursen et al., 2003; MacInnis et al., 2019; Palmer et al., 1996; Smith et al., 2001; Sparks et al., 2016; Sporer & McKenzie, 2007). However, the reliability of cycling performance tests in lesser trained, recreationally active individuals, is unknown. As recreationally active individuals are commonly recruited for sport and exercise science research, determining the reliability of cycling time trial protocols in this population group is of importance. Additionally, the accurate assessment of performance is necessary in strength and conditioning/gymnasium settings to determine the effectiveness of training interventions in clients/athletes. Therefore, the aim of the present study was to investigate the reliability of a self-paced cycling time trial in lesser-trained recreationally active individuals. The main finding was that a 15-minute self-paced cycling time trial was highly reproducible in a cohort of recreationally active individuals.

The high reproducibility of the cycling time trial protocol in this study, demonstrated by a low mean CV of 2.1%, was comparable to other reproducibility studies that have assessed similar duration cycling time trials in trained cyclists (Hickey et al., 1992; Jeukendrup et al., 1996). In a study by Jeukendrup et al. (1996), ten well-trained cyclists completed five trials consisting of a 45-minute preload at 70% W_max followed by a 15-minute time trial. A CV of 3.5% was reported for mean power during the five repeated time trials. Likewise, Che Jusoh et al. (2015) found a CV of 3.6% for work completed during two 15-minute self-paced time trials in the heat. Moreover, Hickey et al. (1992) reported a CV of 0.95% for performance times during four 5-mile time trials (~12 min) in trained cyclists. Similar CV’s, ranging from 0.8% to 3.4%, have been reported for longer duration laboratory-based time trials and when participants have used their own bicycles on air-braked cycle ergometers (Bishop, 1997; Jeukendrup et al., 1996; Laursen et al., 2003; Smith et al., 2001; Sporer & McKenzie, 2007).

To our knowledge, no research has directly assessed the reliability of cycling time trial protocols in recreationally active individuals, however, other studies have reported CVs for multiple familiarisation trials (Cheuvront et al., 2009; Kenefick et al., 2009). These studies used a similar protocol to the current study, both using a 30 minute preload at 50% $\dot{\mathrm{V}}$O_2peak followed by a 15-minute cycling time trial. Cheuvront et al. (2009) reported a CV of 4.5% over three familiarisation time trials, whilst Kenefick et al. (2009) reported a CV of 5% across four familiarisation time trials. However, some of the methods employed in these studies might have reduced the reproducibility of the time trial. Whilst pre-trial diet and exercise routines were standardised in the study of Cheuvront et al. (2009), participants were provided with feedback after each familiarisation trial as motivation to improve for subsequent familiarisation trials. Kenefick et al. (2009) did not standardise diet or activity patterns before the time trials. These methodological considerations might have reduced the reproducibility of the time trial used in these studies. The more stringent standardisation procedures in the present study may well explain the lower CV found in this study compared to these previous studies (Cheuvront et al., 2009; Kenefick et al., 2009). Additionally, both Cheuvront et al. (2009) and Kenefick et al. (2009) used the linear mode on the cycle ergometer during the time trial. The linear mode requires a certain level of cycling experience as the participant must be able to deviate from a selected cadence to adjust their power output. In the present study, the hyperbolic mode was used, allowing participants to directly adjust the workload using the controls on the cycle ergometer monitor, thereby making the time trial cadence independent. Due to this, participants could cycle at a cadence that was comfortable for them, potentially contributing to the lower CV in the present study. Nonetheless, in future research, it would be prudent to measure pedalling cadence, as this could indicate potential alterations in cycling technique or efficiency between experimental conditions that could occur when recruiting recreationally active participants.

The mean CV for the four time trials in the present study was 2.1%, with a range of 0.5–4.9%. When comparing this CV to the CV between the second and third time trials (i.e., one familiarisation trial was completed), there was a small decrease to 1.7% (range 0.2–5.6%). However, when both the first and second trials were considered as familiarisation trials, the mean CV further decreased to 1.0% (range 0.2–1.8%). Similar findings have been reported in well-trained cyclists. Laursen et al. (2003) found that when highly-trained cyclists completed three 40 km cycling time trials on an air-braked cycle ergometer, the CV for performance time decreased from 3.0% to 0.9% when the first trial was excluded as a familiarisation trial. In a shorter duration time trial, Sporer and McKenzie (2007) reported that when trained cyclists completed three 20 km laboratory-based cycling time trials using a Velotron Pro cycle ergometer, the CV for mean power decreased slightly from 2.1% for trials 1 and 2 to 1.9% for trials 2 and 3. In the present study, participants that were reproducible (demonstrated by a CV of <2%) across the first two trials continued to be reproducible across the four trials (Table 2). In contrast, participants that had a CV >2% across the first two trials, all achieved a CV <2% after two familiarisation trials (range 0.2–1.8%). This may be the reason that although a large decrease in mean CV was reported from the first and second time trials to the third and fourth time trials (2.0% to 1.0%), the decrease did not reach statistical significance (P = 0.051). Nonetheless, these results demonstrate that familiarisation is important for the accurate assessment of cycling time trial performance in recreationally active individuals, and that two familiarisation trials may provide more accurate data than one or no familiarisation. It must be noted that although participants were told to complete “as much work as possible” during the time trial, no measurement of confirming maximal effort was taken at the end of the time trials.

The pacing strategies adopted by the participants in this study were similar to those found in other laboratory-based cycling time trials (Corbett et al., 2009; Nikolopoulos et al., 2001; Thomas et al., 2012), with a greater amount of work done at the beginning and end of the time trials, compared to the middle segment. While there were no significant differences in pacing between trials in the present study, there was a degree of variability for the amount of work done at the beginning and the end of the time trials (Figure 1). At the beginning of the time trials there appeared to be a progressive reduction in work done from trial 1 (68.5 ± 13.4 kJ) to trial 4 (66.8 ± 13.3 kJ). On the contrary, there appeared to be a progressive increase in work done from trial 1 (63.0 ± 16.3 kJ) to trial 4 (64.9 ± 15.3 kJ) at the end of the time trials. The alterations in pacing are not surprising due to participants having little or no prior cycling experience, and therefore a lack of familiarity with pacing cycling time trials. However, similar changes in pacing have been reported in well-trained cyclists following repeated time trials (Corbett et al., 2009; Thomas et al., 2012). Thomas et al. (2012) found a trend for a progressively reduced start and progressively increased finish, but no difference in performance times or mean power between trials, when well-trained cyclists completed three repeated 20 km laboratory-based cycling time trials.

The results of this study also have implications for settings outside of laboratory-based experiments, where the ability to understand training responses might be important (i.e., non-cyclist athletes, occupational/military populations, recreational exercisers etc.). Due to the low CV of the performance measure found in the present study, a coach and/or athlete can be confident that a short-duration self-paced time trial, which is relatively simple to implement in an applied setting (e.g., in a gymnasium etc.), can be used to determine the effectiveness of a training intervention and/or monitor performance throughout an athlete’s training period. The increased popularity of mass participation endurance events, such as cycling sportives, 5 km and 10 km running events, means quantifying the performance of lesser trained populations can facilitate optimisation of their performance in such events, and can provide a measure for personal trainers, coaches and strength and conditioning practitioners to support training prescription.

In conclusion, the results from the present study demonstrate that the use of a 15-minute self-paced cycling time trial in recreationally active individuals is a reliable performance measure. Reliability of the performance measure improved with familiarisation, particularly in those with poor test-retest reliability over the first two time trials. Two familiarisation trials ensured all participants had a test-retest reliability of <2%. Future research should seek to examine the pacing strategies adopted by this population, as well as reliability of longer duration cycling time trial protocols.

Disclosure statement

MPF has no conflicts of interest. LJJ is part of the National Institute for Health Research’s Leicester Biomedical Research Centre, which is a partnership between University Hospitals of Leicester NHS Trust, Loughborough University, and the University of Leicester. This report is independent research by the National Institute for Health Research. The views expressed in this publication are those of the authors and not necessarily those of the NHS, the National Institute for Health Research, or the Department of Health. LJJ has current/previous funding from Entrinsic Beverage Company LLC, Entrinsic Bioscience LLC, Herbalife Europe Ltd, Bridge Farm Nurseries, Decathlon SA, PepsiCo Inc., Volac International; has performed consultancy for PepsiCo Inc. and Lucozade, Ribena Suntory; and has received conference fees from PepsiCo Inc. and Danone Nutricia. In all cases, monies have been paid to LJJs institution and not directly to LJJ. SAM has current/previous funding from Entrinsic Beverage Company LLP and Herbalife Europe Ltd.

Data availability statement

Data generated or analysed during this study are available from the corresponding author upon reasonable request.

Correction Statement

This article has been corrected with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.