The effect of running versus cycling high-intensity intermittent exercise on local tissue oxygenation and perceived enjoyment in 18–30-year-old sedentary men

Ethics statement

This research study was approved by the human research ethics committee of the University of the Sunshine Coast (S/13/472). All participants received a research study information sheet before providing written informed consent.

Experiment design

The study consisted of three testing sessions: a maximal incremental exercise test conducted on a treadmill ergometer (MAX) followed by two experiment conditions: (1) a protocol of free-paced HIIT conducted on a bicycle ergometer (HIITCYC); and (2) a protocol of constant-paced HIIT conducted on a treadmill ergometer (HIITRUN). The HIIT conditions were randomized to control for any possible order effect. All testing sessions were separated by three to seven days to minimise the influence of any potential carry-over effects between testing sessions. The ΔHHb in the gastrocnemius (GN), the left vastus lateralis (LVL) and the right vastus lateralis (RVL) muscles, VO₂, HR, RPE and the PACES were measured during the HIIT sessions. The format of the HIIT testing sessions and the timing of measurements are illustrated in Fig. 1.

Participants

The participant group consisted of 12 men who met the inclusion criteria of being aged 18–30 years; currently completing less than 150 min of moderate intensity or 75 min of vigorous intensity activity per week; reporting no cardiovascular and metabolic disease; taking no medications; having no known health-related issues that would be made worse by, or inhibit participation in the study. Descriptive physical characteristics of participants are presented in Table 1.

Screening procedures

At the initial session, participants completed risk screening questionnaires, a physical activity log and physical characteristics of height, mass, resting pulmonary function and adipose tissue thickness (ATT) were measured (Table 1), as previously described (Kriel et al., 2016). Briefly, the physical activity log was used to ensure that participants’ activity levels over the previous three months met the definition of sedentary for the purposes of this study (i.e., not achieving the current minimal recommendations for exercise participation to obtain health benefits) (Barnes et al., 2012). Participants were asked to not perform exercise in the 24 h prior to each session and to not consume any caffeine, alcohol or a large meal in the 4 h preceding each session (Goldstein et al., 2010). It was confirmed at each testing session that participants were appropriately fed and hydrated, in line with existing pre-exercise nutrition and hydration guidelines (Convertino et al., 1996; Garzon & Mohr, 2014).

Exercise sessions

Prior to the initial exercise session, participants were familiarised with the testing protocols, the cycle ergometer (Veletron; Racermate, Seattle, WA, USA) and the treadmill (T200; Cosmed, Rome, Italy). A safety harness was worn during all treadmill sessions.

Each exercise session began with a 3 min baseline data collection period during which the participant either sat on the cycle ergometer or stood on the treadmill. The baseline period was followed by a 4 min warm up period, consisting of either cycling against a constant resistance of 60 Watts (W) (HIITCYC) or walking at a 10% gradient and an individually determined speed calculated to produce approximately 60 W of resistance (HIITRUN), calculated from the rate of vertical displacement (running speed and gradient) and body mass.

During the maximal incremental exercise test, the warm up was followed by an incremental speed protocol at a fixed 10% gradient, with an initial running speed of 8 km·h⁻¹, increasing by 1 km·h⁻¹ every 30 s until volitional cessation. The peak treadmill running speed achieved during MAX was used as the participants’ running speed during the HIITRUN bouts. The 10% grade was chosen to simulate individuals performing HIIT by running up a steep incline, to enable mechanical power to be calculated and to provide an intensity of exercise that would produce a maximal sprint effort without relying primarily on running speed. It was necessary to determine a realistic maximal running speed for the 30 s HIITRUN bouts, which would potentially not have been achieved with longer stage durations in untrained sedentary participants during the maximal incremental test, due to fatigue-inducing exercise cessation prior to the attainment of maximal running speed (Kang et al., 2001; Kirkeberg et al., 2011).

The format of the HIIT experiment conditions, as illustrated in Fig. 1, have been described previously (Kriel et al., 2016). Briefly, the warm up was followed by four 30 s bouts of HIIT, with 2 min passive recovery periods separating each of the HIIT bouts. The four bouts of HIIT during HIITCYC consisted of repeat Wingate sprints. The four bouts of HIIT during HIITRUN consisted of participants running at the maximal speed achieved during the maximal incremental test (MAX), for each of the four bouts. During the passive recovery periods, participants were instructed to sit (HIITCYC) or stand (HIITRUN) as still as possible to reduce movement artefact in the NIRS data. Following the final bout, there was a 6 min passive recovery period. The time-efficient HIIT protocol format was based upon protocols used in trained and untrained populations (Buchheit et al., 2010, 2012; Burgomaster et al., 2005, 2008; Dupont et al., 2007; Gibala et al., 2006; Gibala & McGee, 2008).

Tissue oxygenation

Changes in local tissue oxygenation were measured continuously, as illustrated in Fig. 1, and as described previously (Kriel et al., 2016). Briefly, the changes in the relative concentration of HHb (Δ[HHb]) were measured using a NIRS system (three x PortaMon devices; Artinis Medical Systems BV, Zetten, The Netherlands). The PortaMon devices were placed on the skin directly over the muscle belly of three locomotor muscles: the LVL, the RVL and the left GN. Test-retest reliability of the HHb data from the NIRS system was examined prior to this study, providing acceptable absolute reliability values of the baseline data for each site (Typical Error: VL = 0.4 μM, GN = 0.8 μM) (Kriel et al., 2016). Furthermore, previous research has shown that the NIRS method provides acceptable reliability for the measurement of HHb in active muscle tissue (Austin et al., 2005; Muthalib et al., 2010). For all testing, the same device was used at the same measurement site. To ensure placement consistency, positioning of the individually labelled NIRS devices was referenced to anatomical landmarks as detailed previously (Buchheit et al., 2009, 2011; Dupont et al., 2007; Prieur & Mucci, 2013; Smith & Billaut, 2010). The location of devices was marked with an ink pen at the first session and participants were required to maintain the marks between subsequent sessions.

For this study, only Δ[HHb] values are presented, as discussed previously (Kriel et al., 2016). Briefly, the Δ[HHb] data are potentially unaffected by changes in perfusion, blood volume or arterial haemoglobin concentration (Adami et al., 2015; Jones et al., 2009; Wang et al., 2006). Movement artefact was present in NIRS data collected at the LVL, RVL and GN sites during the passive recovery periods of the HIIT sessions, attributed to a low signal to noise ratio. Therefore, recovery period data were not included in further analysis. Similar technical difficulties in NIRS data collection, leading to data exclusion, have been reported (Mekari et al., 2015). The inter-individual variability in the Δ[HHb] data (range −5.77 to 32.33 μM during the HIIT bouts) is presented in Fig. 2 for the RVL during HIITCYC.

Systemic oxygen consumption, HR and mechanical power

Systemic oxygen consumption (VO₂) data were collected continuously, as illustrated in Fig. 1, using a respiratory gas analysis open circuit spirometry system (Parvo Medics, Sandy, UT, USA). Standardised calibration and methods were used (Macfarlane & Wu, 2013). A HR monitor (RS400; Polar Electro, Kempele, Finland) was used to measure and record HR, as illustrated in Fig. 1. A crank-based power meter (SRM Science; Schoberer Rad Meßtechnik, Julich, Germany) was used during HIITCYC to measure and record mechanical power output. The power output during the HIITRUN bouts was a product of the constant speed, the fixed 10% gradient and the participant’s body weight using the equations: Power = Work/Time; Work = body weight × total vertical distance (speed × gradient × time), used previously when comparing high-intensity bouts of uphill running to non-steady state cycling (Scott et al., 2006).

Ratings of perceived exertion

To determine perceived exertion, participants were provided with a standardised description of the CR10 RPE scale (Borg, 1982) and the scale’s purpose, including memory anchoring of the scale (an explanation of the sensations associated with the high and low scale categories) (Gearhart et al., 2004). Participants were asked to provide an RPE score immediately after each bout of HIIT as well as give a ‘Condition’ RPE score (RPE^COND) 1 min after the final (fourth) bout. This RPE^COND score was used to compare the overall perception of exertion between the HIIT conditions.

Physical activity enjoyment scale

To determine enjoyment levels in response to each condition, participants completed the PACES questionnaire within 5 min of completing each condition. The PACES was used to compare enjoyment levels between the HIIT conditions. The PACES consists of eighteen items on a seven-point bipolar scale. A minimum total score of 18 and a maximal total score of 126 is possible. The PACES is a reliable and valid measure of enjoyment during HIIT (Oliveira et al., 2013; Tritter et al., 2013).

Data calculation

All NIRS data were collected at 10 Hz, smoothed using a 10-point moving average and then averaged to 1 s periods for statistical analysis. The NIRS data were expressed as units of change (μM) from the mean value of the 30 s of baseline data preceding the start of exercise (Δ[HHb]). VO₂ data were averaged over 5 s periods while HR and HIITCYC power data were averaged at 1 s intervals initially. The NIRS, HR, VO₂ and HIITCYC power data were then time aligned. The time periods of data corresponding to the four 30 s bouts of HIIT were then identified. Mean 30 s values were then calculated for all dependant variables for each bout of HIIT, providing a single value per bout for statistical analysis. RPE and PACES data provided a single value per measurement time for statistical analysis.

Statistics

Statistical tests were performed using the IBM SPSS Statistics (version 22; IBM Corporation, Armonk, NY, USA) program. Data were initially tested for normality of distribution using the Shapiro–Wilk test. A three factor, repeated-measures analysis of variance (ANOVA) was used to analyse the effect of condition, bout and site on the dependant variable: Δ[HHb]. A two factor, repeated-measures ANOVA was used to analyse the effect of condition and bout on the dependant variables of VO₂, HR, mechanical power and RPE. If a significant main effect or interaction effect was identified, a Bonferroni’s post hoc test was used to make pair wise comparisons. A paired samples T-test was used to analyse the effect of condition on PACES scores. All variables are presented as mean ± SD. For all statistical analyses, a p value of < 0.05 was used as the level of significance. Effect size estimates are indicated using partial η² values.

Tissue oxygenation

For the mean Δ[HHb] for all sites, conditions and bouts combined, there was a main effect for site (p = 0.002, F = 6.057, η_p² = 0.377), condition (p < 0.001, F = 43.392, η_p² = 0.813) and bout (p = 0.022, F = 5.641, η_p² = 0.361). There were condition × site (p < 0.001, F = 8.031, η_p² = 0.445), condition × bout (p = 0.002, F = 9.643, η_p² = 0.491) and site × bout (p < 0.001, F = 7.615, η_p² = 0.432) interactions. For the mean Δ[HHb] for each condition, there was a main effect for site for HIITCYC (p = 0.001, F = 6.846, η_p² = 0.384) and HIITRUN [(p < 0.001, F = 8.981, η_p² = 0.473). There was a main effect for bout (p < 0.001, F = 15.849, η_p² = 0.590) and site × bout interactions (p = 0.003, F = 5.353, η_p² = 0.327) for HIITCYC.

For the GN no significant differences were found for condition (p = 0.685) or bout (p = 0.057) (Fig. 3A).

For the LVL, there was a main effect in the mean Δ[HHb] for condition: [(p = 0.001, F = 17.647, η_p² = 0.616) HIITCYC 10.74 ± 8.53 μM; HIITRUN 1.87 ± 3.43 μM]. There was no significant condition × bout interaction. For the mean Δ[HHb] for each bout (Fig. 3B), differences were found between conditions for Bout 1 (p = 0.002, F = 17.344, η_p² = 0.612), Bout 2 (p = 0.003, F = 14.593, η_p² = 0.57), Bout 3 (p = 0.001, F = 17.897, η_p² = 0.619) and Bout 4 (p = 0.001, F = 18.087, η_p² = 0.622) with Δ[HHb] higher during HIITCYC, compared to HIITRUN. For the mean Δ[HHb] within conditions, a significant increase was found when comparing Bout 1–Bout 3 in the HIITCYC condition only (p = 0.006, F = 7.994, η_p² = 0.421).

For the RVL, there was a main effect in the mean Δ[HHb] for condition: [(p = 0.002, F = 16.347, η_p² = 0.598) HIITCYC 11.16 ± 7.99 μM; HIITRUN 3.51 ± 4.88 μM]. There was a condition × bout interaction (p = 0.045). For the mean Δ[HHb] for each bout (Fig. 3C), differences were found between conditions for Bout 1 (p = 0.002, F = 15.371, η_p² = 0.583), Bout 2 (p = 0.003, F = 13.762, η_p² = 0.556), Bout 3 (p = 0.002, F = 16.362, η_p² = 0.598) and Bout 4 (p = 0.001, F = 18.134, η_p² = 0.622) with Δ[HHb] higher during HIITCYC, compared to HIITRUN. For the mean Δ[HHb] within conditions, there were no significant differences found across bouts.

Systemic oxygen consumption

For the VO₂, there was a main effect for condition (p = 0.017, F = 9.118, η_p² = 0.533) and bout (p = 0.008, F = 5.016, η_p² = 0.385). There was a condition × bout interaction (p = 0.021, F = 3.885, η_p² = 0.327).

For the mean VO₂ for each bout (Fig. 4A), differences were found between conditions for Bout 2 (p = 0.003, F = 14.807, η_p² = 0.597) and Bout 3 (p = 0.012, F = 9.317, η_p² = 0.482) with VO₂ higher during HIITCYC, compared to HIITRUN. For the mean VO₂ within conditions, values increased over time in HIITCYC (p = 0.016, F = 6.347, η_p² = 0.414) when comparing Bout 1 to Bout 2 and Bout 3.

Heart rate

For the HR, there was a main effect for condition (p < 0.001, F = 31.126, η_p² = 0.757). There was a condition × bout interaction (p = 0.014).

For the mean HR for each bout (Fig. 4B), differences were found between conditions for Bout 2 (p < 0.001, F = 26.002, η_p² = 0.722), Bout 3 (p < 0.001 F = 53.022, η_p² = 0.841) and Bout 4 (p < 0.001, F = 37.521, η_p² = 0.79) with HR higher during HIITCYC, compared to HIITRUN. For the mean HR within conditions, values increased over time in HIITCYC (p < 0.002, F = 12.316, η_p² = 0.804) when comparing Bout 1 to Bout 3 and 4.

Mechanical power

For the power output, there was a main effect for bout (p = 0.001, F = 11.205, η_p² = 0.528). There was a condition × bout interaction (p = 0.001).

For the mean power output for each bout (Fig. 4C), differences were found between conditions for Bout 4 (p = 0.008, F = 10.728, η_p² = 0.518) with mechanical power lower during HIITCYC, compared to HIITRUN. For the mean power output within conditions, values decreased over time in HIITCYC (p = 0.001, F = 11.205, η_p² = 0.528) when comparing Bout 1 and 2 to Bout 3 and 4.

Ratings of perceived exertion

For the mean RPE, there was a main effect for condition (p < 0.001, F = 79.976, η_p² = 0.885) and bout (p = 0.001, F = 14.259, η_p² = 0.765). There was no significant condition × bout interaction.

For the mean RPE for each bout (Fig. 5A), differences were found for Bout 1 (p = 0.001, F = 20.477, η_p² = 0.651), Bout 2 (p < 0.001, F = 95.703, η_p² = 0.905), Bout 3 (p < 0.001, F = 70.304, η_p² = 0.865) and Bout 4 (p < 0.001, F = 85.105, η_p² = 0.886) with RPE higher during HIITCYC, compared to HIITRUN. For the mean RPE within conditions, values increased over time in the HIITCYC (p < 0.001, F = 44.704, η_p² = 0.803) and HIITRUN (p < 0.017, F = 6.251, η_p² = 0.405) conditions when comparing Bout 1 to Bout 2, 3 and 4 during HIITCYC and Bout 1, 2 and 3 to Bout 4 during HIITRUN.

The RPE^COND score was higher for HIITCYC, compared to HIITRUN, (p < 0.001, F = 37.027, η_p² = 0.771) (Fig. 5B).

Physical activity enjoyment scale

The PACES score was found to be higher for HIITRUN when compared to HIITCYC; t (11) = −2.460, p = 0.032, η_p² = 0.355 (Fig. 5C).

Tissue oxygenation

A significantly higher Δ[HHb] at the LVL and RVL sites occurred during all bouts of HIITCYC, compared to HIITRUN. This is indicative of increased local oxygen utilisation in the large locomotor muscles during HIITCYC which is consistent with a higher systemic oxygen utilisation, as indicated by higher VO₂ during HIITCYC. It is conceivable that in a population with no specificity of training in cycling, during high-intensity exercise, a higher local oxygen utilisation (linked to metabolic demand) will exist during HIITCYC, compared to HIITRUN. This is irrespective of the fact that during incremental and steady-state exercise vastus lateralis muscle activity has been found to be greater or equal during running, compared to cycling (Bijker, de Groot & Hollander, 2002; Bouillon et al., 2016; Millet & Lepers, 2004).

In the smaller GN muscle of the leg, which generally has a greater percentage of oxidative muscle fibres (Hagström-Toft et al., 2002; Houmard et al., 1998) and greater citrate synthase activity (Houmard et al., 1998) than the vastus lateralis muscle, there were no significant differences in Δ[HHb] between conditions or bouts. In muscles comprised of oxidative fibres, an improved matching of oxygen supply and demand has been shown, when compared to muscles comprised of glycolytic fibres (Boone et al., 2016). This indicates that during exercise, GN was able to meet the local increased energy requirements, creating little change in HHb values compared to pre-exercise values, irrespective of the mode specific differences in systemic and vastus lateralis oxygen utilisation. Additionally, while GN muscle activation is generally higher during cycling than running at the same HR (Bouillon et al., 2016), it has been shown that running on an incline leads to a greater activation of the GN than during level running (Millet, Vleck & Bentley, 2009). Therefore, it is conceivable that the greater activation of GN during the incline running protocol of HIITRUN resulted in a similar level of oxygen utilisation in GN as the cycling protocol during HIITCYC.

When comparing Δ[HHb] within conditions, there was no overall progressive increase in muscle oxygen utilisation in GN, LVL and RVL from Bout 1 to Bout 4 irrespective of exercise mode. Exercise effort was submaximal (as indicated by VO₂, HR and RPE data) and if the 2 min passive recovery periods allowed adequate reoxygenation and recovery in the vasculature of the working muscles, the muscles would be able to meet the energy requirements of each bout with existing oxygen supply, creating little change in HHb values across bouts. The lack of significant change in Δ[HHb] at the locomotor muscle sites from Bout 1 to 4 indicates that the level of maximal oxygen utilisation achieved during the conditions was not increased by the addition of multiple exercise bouts, irrespective of the mode of exercise adopted. This suggests that a single bout may be sufficient to induce maximal levels of oxygen utilisation, and this has implications for identifying the smallest dose of HIIT needed to convey the health benefits attributed to this format of exercise.

The inter-individual variability in the RVL Δ[HHb] response during the HIITCYC condition, as illustrated in Fig. 2, was unrelated to the participants power output (i.e. the participants with the highest power outputs did not show the greatest increases in HHb). A similar level of variability in the Δ[HHb] response can be seen in the two projects to publish individual results (Jones, Hamilton & Cooper, 2015; Kriel et al., 2016). While recommendations for performing muscle NIRS measurements and NIRS data analysis are available (Ferrari, Muthalib & Quaresima, 2011; Grassi & Quaresima, 2016; Perrey & Ferrari, 2018), there is no accepted standard for the method of calculating, analysing and presenting individual and group NIRS data. Furthermore, factors affecting the quality and integrity of NIRS data (such as ATT and optode positioning) are often not reported (Perrey & Ferrari, 2018). This makes comparison of Δ[HHb] data between projects difficult, even when projects have been performed in similar populations, performing similar HIIT interventions.

To our knowledge, this is the first study to show that in sedentary participants, the Δ[HHb] levels attained during HIIT varies in three locomotor muscles of the lower limbs, due to the mode of HIIT exercise. This finding indicates the specificity of oxygen utilisation in the sedentary population.

Systemic oxygen consumption

Higher peak VO₂ and HR values are frequently reported during incremental running exercise, compared to incremental cycling exercise, due to the increased metabolic cost of weight bearing ambulation (Abrantes et al., 2012; Basset & Boulay, 2000; Hill, Halcomb & Stevens, 2003; Scott et al., 2006).

In contrast, the VO₂ and HR responses were not different at Bout 1 between HIITCYC and HIITRUN, indicating that initially the conditions were well matched for intensity. Furthermore, significantly higher VO₂ during Bout 2 and Bout 3 of HIITCYC, compared to HIITRUN, indicates an increased exercise intensity and integrated physiological demand during the non-weight bearing cycling condition, contributed to by an increased local oxygen utilisation, as evidenced by the higher Δ[HHb] discussed previously. In a sedentary population, it is assumed that at least a partial application of the specificity of training principle will exist for walking/running (due to everyday ambulation) while no such training effect would be present for cycling (no participants reported routine cycling activity when completing their physical activity log). This lack of adaptation, both centrally and peripherally, could have contributed to the increased intensity and physiological demand during HIITCYC.

When examining VO₂ changes within each condition, a potential explanation for there being no difference in VO₂ between Bout 2, 3 and 4 of HIITCYC could be the significant decrease in mechanical power output across bouts, effectively requiring less aerobic contribution over time, in effect counteracting the expected continued increase in VO₂ in later bouts due to the cumulative load of the protocol. Exercise effort was submaximal during HIITRUN (as indicated by VO₂, HR and RPE data) and it is conceivable that the 2 min passive recovery periods allowed adequate recovery from this submaximal exercise. Therefore, VO₂ remained relatively stable over time.

The sedentary participants were potentially limited in their ability to achieve true maximal running speeds during the incremental treadmill test (MAX) due to possible slower VO₂ kinetics, poor running efficiency, and lack of familiarity with running at rapidly increasing speeds at a 10% gradient. This is however an important observation: it appears that the (in)ability to run at speeds associated with maximal/supramaximal exercise may limit the exercise intensity able to be achieved by sedentary individuals during short duration treadmill running HIIT when compared to short duration cycling HIIT. A potential future research direction may be to compare the beneficial physiological and perceptual effects of less time-efficient, but submaximal HIIT protocols performed on the treadmill, such as the 4 × 4 min protocol popular in clinical research (Ramos et al., 2016), to Wingate cycling HIIT in sedentary participants.

Heart rate

The significantly higher mean HR during Bout 2, 3 and 4 of HIITCYC, compared to HIITRUN, are further indications of an increased exercise intensity and physiological demand during the cycling condition.

The relatively static HR response over time, when comparing Bout 2, 3 and 4 of HIITCYC, supports similar findings in which a passive recovery HIIT protocol was used (Lopez, Smoliga & Zavorsky, 2014). In the HIITRUN condition, similar to VO₂ values, exercise effort was submaximal and the 2 min recovery periods allowed adequate recovery, it is conceivable that HRs would also remain relatively stable over time.

Mechanical power

The most frequently used modes of HIIT are running and cycling, utilising ergometers (Kessler, Sisson & Short, 2012; Logan et al., 2014; Weston et al., 2014). However, as mentioned previously, there are differences in how running and cycling ergometers are utilised to induce the requisite physiological stress. Cycle ergometer HIIT, during which participants do not support their own body weight, safely lends itself to all-out efforts, and this is typically achieved (Weston et al., 2014). During treadmill ergometer HIIT, utilising motorised treadmills, the ability to safely produce short duration all-out efforts is limited. Typically, a prior indication of maximal running speed is required so that the speed (and gradient) of the HIIT session can be set to produce a high-intensity effort (Ben Abderrahman et al., 2013; Iaia et al., 2009). Therefore, the mechanical power produced during HIITRUN and HIITCYC were not directly matched. However, when comparing mechanical power between conditions, no difference in mechanical power output occurred until Bout 4, when the mechanical power in HIITCYC was lower than in HIITRUN. The late reduction in mechanical power during HIITCYC, compared to HIITRUN, could be attributed to a greater cumulative fatigue due to the all-out format of the cycling bouts.

Power output (and exercise intensity) remained stable and submaximal for HIITRUN from Bout 1 to Bout 4. The submaximal intensity potentially enabled a more consistent effort during exercise, and enhanced recovery, during HIITRUN, hence potentially less cumulative fatigue developed. The higher relative intensity of the all-out cycling bouts during HIITCYC potentially lead to the significant decline in power output from Bout 1 to Bout 4 and the higher Δ[HHb], VO₂ and HR responses (Dupont et al., 2007; Lopez, Smoliga & Zavorsky, 2014; Wahl et al., 2013). This is further evidence of cumulative fatigue during the cycling condition and was expected due to the incomplete ATP repletion and phosphocreatine resynthesis that occur during repeat Wingate exercise that do not allow time for complete recovery (Dupont et al., 2007; Lopez, Smoliga & Zavorsky, 2014; Wahl et al., 2013). The differences in mechanical power output and relative intensity between HIITRUN and HIITCYC is an important finding, demonstrating that when utilising typical HIIT protocols available to sedentary individuals looking to initiate physical activity in recreational exercise settings, cycling is likely to result in greater perturbations in physiological responses compared to running.

While the aim of this study was not to match the exercise intensity of running and cycling HIIT sessions, future research in which absolute workload of the HIIT session is matched would assist in determining to what extent the differences in responses observed in this study were the result of mode and/or intensity.

Ratings of perceived exertion

Ratings of perceived exertion was higher during the first bout of HIITCYC, compared to HIITRUN, despite no difference in markers of physiological intensity (VO₂ and HR) or mechanical power. This finding is consistent with previous research indicating that exercise performed at the same absolute intensity corresponds to a higher relative intensity in cycling (Abrantes et al., 2012). RPE was also higher during Bouts 2–4 of HIITCYC compared to HIITRUN. This finding is consistent with previous research which incorporated incremental (Abrantes et al., 2012) or steady state (Thomas et al., 1995) exercise. RPE is a subjective indicator of intensity during exercise (Pescatello & American College of Sports Medicine, 2013) and taken together with the higher Δ[HHb], VO₂ and HR response during Bouts 2–4 of HIITCYC provide further evidence of a higher physiological strain, a higher level of cumulative fatigue during HIITCYC and a potential lack of habitual adaptation to cycling.

Physical activity enjoyment scale

The examination of sedentary individuals’ enjoyment during exercise has provided inconsistent results: Sedentary individuals are more likely to enjoy and be compliant with exercise that they perceive as moderate in intensity (Williams et al., 2008), however HIIT has also been shown to be enjoyable and elicit a degree of positive affect (Martinez et al., 2015). Exercise mode has been shown to have an acute effect on psychological mood state, including enjoyment, with running eliciting a more positive mood profile than weightlifting (Dyer & Crouch, 1988). However, the effect of mode on sedentary individuals’ enjoyment levels during HIIT is unknown.

Participants enjoyed the HIITRUN session more. When taking into account the significantly lower RPE scores in HIITRUN when compared to HIITCYC, this finding is in agreement with research that has found that exercise that elicits a lower RPE score is more enjoyable (Kilpatrick et al., 2003).

Adverse reactions to HIIT exercise are rarely reported (Weston, Wisloff & Coombes, 2014). During the current study, seven participants reported either leg muscle discomfort/cramping (n = 3) or dizziness and nausea (n = 4) during the recovery period of the HIITCYC condition. Although all participants felt well upon leaving the laboratory (no more than 30 min post exercise), these physiological responses could have contributed to the lower levels of enjoyment during HIITCYC.

Limitations

Up to nine design variables can be adjusted when designing HIIT protocols (Buchheit & Laursen, 2013b), leading to large variability in HIIT protocol composition within the literature. The 30 s bouts and 2 min recovery periods (a 1:4 work to recovery ratio) limits the generalisability and comparison of the research findings to the broader HIIT literature.

It is routine to wait ≈15 min to collect an ‘overall RPE’ after HIIT, due to the discomfort of the last bout potentially affecting the overall figure. However, adverse reactions experienced by seven participants during this study were noted in previous studies and generally occurred during the latter half of the 6 min post-exercise recovery period. Collecting the RPE^COND score earlier in recovery, while potentially introducing an exercise related bias and therefore a potential limitation to this study, avoided a potential bias due to adverse sensations experienced later in recovery.

The PACES data were collected retrospectively, following exercise. The PACES score was likely to be influenced by the participants’ physiological and psychological state at that time. This study would perhaps have been improved by assessing enjoyment during exercise, however this was deemed impractical due to experimental design and physiological data collection methods.

It is acknowledged that sedentary participants can exhibit slower VO₂ kinetics, hence VO_2max values may have been underestimated by using stages of 30 s duration during the maximal incremental exercise test, compared to standard stages of 3 min duration. However, no difference in VO_2max has been shown with short versus traditional length protocols (Kang et al., 2001; Midgley et al., 2008). Furthermore, a VO_2max of 43.5 ± 4.3 ml·kg·min⁻¹, classified as the 35th percentile in terms of fitness (American College of Sports Medicine et al., 2018; Kaminsky, Arena & Myers, 2015), is relatively high for sedentary participants, making underestimation of VO_2max unlikely given that the reported mean physical activity time of a moderate intensity for this group was 31 ± 33 min·week⁻¹.