Abstract
The metabolic cost of steady-state walking is well known; however, across legged animals, most walking bouts are too short to reach steady state. Here, we investigate how bout duration affects the metabolic cost of human walking with varying mechanical power, metabolic intensity and duration. Ten participants walked for 10- to 240-s bouts on a stair climber at 0.20, 0.25 and 0.36 m s−1 and on a treadmill at 1.39 m s−1. Oxygen uptake was time-integrated and divided by bout duration to get bout average uptake (V̇O2(b)). Fitting of oxygen uptake kinetics allowed calculating non-metabolic oxygen exchange during phase-I transient and, hence, non-steady-state metabolic cost (C met(b)) and efficiency. For 240-s bouts, such variables were also calculated at steady state. Across all conditions, shorter bouts had higher V̇O2(b) and C met(b), with proportionally greater non-metabolic oxygen exchange. As the bout duration increased, V̇O2(b), C met(b) and efficiency approached steady-state values. Our findings show that the time-averaged oxygen uptake and metabolic cost are greater for shorter than longer bouts: 30-s bouts consume 20–60% more oxygen than steady-state extrapolations. This is partially explained by the proportionally greater non-metabolic oxygen uptake and leads to lower efficiency for shorter bouts. Inferring metabolic cost from steady state substantially underestimates energy expenditure for short bouts.
1. Introduction
The energy demands of locomotion have been extensively studied at metabolic steady state [1–3], which requires several minutes to be attained. However, most locomotion bouts are brief, consisting of a few consecutive steps [4]: quantifying their energy expenditure is pivotal to shed light on how animals allocate their energy budgets, choose gaits, speeds and paths that optimize metabolic costs [5–8] or trade-off metabolic cost with time [9–11], muscle activity [12] and stability [13]. For humans, integrating brief bouts of walking into the daily routine is also a cornerstone in promoting physical activity and health [14–21]. Exercise programmes based on short walking bouts are used for the treatment and rehabilitation of several clinical conditions, including obesity and stroke [22,23]. Understanding the cost of short bouts is crucial for tailoring them, especially for people with low aerobic fitness and increased time to reach a metabolic steady state [24–26].
Determining the energy demands of locomotion at non-steady-state is not trivial. Classical indirect calorimetry assumes proportionality between the rate of recovery reactions replenishing high-energy phosphate bonds during exercise and the alveolar oxygen uptake (V̇O2), with minimal anaerobic contributions [27,28]. However, at the onset of the bout, energy is generated mainly through anaerobic processes. Anaerobic metabolism at bout start has been traditionally assessed based on the concept of oxygen deficit-debt [29–33]: if all the oxygen stores and metabolites return to resting values during recovery, the summed excess—or above resting—exercise (excess exercise oxygen consumption, EEOC) and post-exercise (excess post-exercise oxygen consumption, EPOC) oxygen volume should be proportional to the energy produced through the summed anaerobic and aerobic mechanisms. In fact, however, the summed EEOC and EPOC exceed what would be predicted by steady-state oxygen uptake measures: such a difference appears consistently present across gaits, developmental stages and species, and increases with decreasing bout duration [9,11,34–40]. Nonetheless, it remains challenging to understand the mechanisms underlying these discrepancies and to reconcile exercise oxygen consumption with accurate quantification of energy expenditure during locomotion.
From a metabolic perspective, it could be questioned whether all EEOC serves metabolic purposes. During the V̇O2 on-transient, at least two distinct responses are observed: an immediate phase I and a delayed phase II. While the latter matches oxygen uptake at the muscles [41], the former is attributed to cardiodynamic processes [42], with additional oxygen uptake that does not reflect muscle oxygen demand [43]. The corresponding oxygen volume may hence be subtracted from the EEOC when assessing the bioenergetics of short bouts of locomotion. A second assumption is that all EPOC is allocated to metabolic processes. In the decades following the works of Hill et al. and Margaria et al. [30,32], researchers have highlighted other mechanisms influencing EPOC, including changes in circulating hormones, body temperature, ventilation and substrate metabolism [44–48]. Such mechanisms may fall into two categories: those increasing adenosine triphosphate (ATP)-dependent processes other than actomyosin cross-bridge cycling (e.g. increased active ion permeability) [45,46] and those replacing oxygen stores or changing the stoichiometry between oxygen consumption and ATP production [45,47,48]. The former mechanisms should be considered in the energy spent to do the locomotor bout because they are caused by the bout itself, and may impact shorter bouts proportionally more than longer ones. The latter [27,45] influence the observed ratio between energy yield and oxygen consumption (energy equivalent of oxygen, EqO2) and should be corrected for when doing indirect calorimetry for short locomotion bouts [49–52]. The EqO2 may indeed be higher during the on-transient than at steady state and lower during the recovery [49,50]. Finally, the distinction between EEOC and EPOC is not straightforward. Indeed, the limb-to-lung transit time can be of the same order of magnitude as the duration of the short bouts themselves [53], and literature suggests that for such short bouts, V̇O2 could actually continue to increase after the bout has ended [39,54,55].
From a mechanical perspective, it should be assessed whether efficiency—the ratio of mechanical to metabolic work [56]—changes with bout duration. Studies on mice, lizards and humans suggest that short bouts of walking are less efficient than longer ones [9,11,35,39], which agrees only partially with evidence from isolated limb exercises and isolated muscular contraction [57–61]. Efficiency should, however, be calculated after correcting for the mentioned metabolic determinants and by standardizing mechanical work [62].
In this context, two activities appear as suitable models to study short bouts of locomotion and daily energy budgets in humans: walking up a stair treadmill and walking on a flat treadmill. The former mainly requires positive mechanical work to move, with minimal contribution from elastic storage and negative work [63], and allows controlling for stride frequency, which may confound internal work and metabolic demands [64–66]. For such an activity, data from Minetti et al. [39] suggest a difference in oxygen demands at varying bout durations, but lack of standardization of mechanical power and duration allows limited inference. Flat walking has a lower intensity, reducing potential confounding factors owing to altered oxygen kinetics in higher-intensity exercise domains [53]; also, for this activity, Blokland et al. [35] reported differences in oxygen consumption between 1-, 2- and 6-min bouts. Hence, the aim of this study was to evaluate how bout duration affects the metabolic cost, mechanical demands and efficiency of stair climbing and level walking in humans under standardized bout durations and mechanical work.
2. Material and methods
(a) Experiment 1: stair climbing
Ten healthy young participants (five males and five females; age 27 ± 5 years, body mass 69.5 ± 14.1 kg, height 1.74 ± 0.10 m; physical activity: 1890 ± 752 metabolic equivalent task (MET)-min week−1) without disorders impacting gait were recruited for the study. All participants gave informed written consent after being informed about the study, in accordance with the Declaration of Helsinki (except for registration in a database) and with the permission of the Ethics Committee of the University of Milan. Participants walked on a stair climber (Climb Excite; Technogym, Cesena, Italy) with a 78% incline (figure 1a ) at diagonal speeds of 0.20, 0.25 and 0.36 m s−1 while their gas exchanges were measured by a portable metabolic system (mass: 0.9 kg, K5; Cosmed, Rome). For each speed, participants performed bouts lasting 10, 30, 60, 90 and 240 s, each one repeated twice, in a randomized order. Before each bout, resting V̇O2 was measured for 3 min with participants sitting on a chair; following an auditory signal, participants then started walking on the stair climber belt, which was already at target speed. After each bout, V̇O2 was measured during the recovery for 7 min while participants were sitting on the same chair (figure 1b ). The duration of 7 min was chosen to keep the measurement bias for oxygen volume under 5% of predicted steady-state measures (electronic supplementary material, S1). Additionally, four complete V̇O2 kinetics were collected at 0.36 m s−1 for bouts lasting 240 s. For each participant, acquisitions were broken down into 7–9 experimental sessions on separate days to avoid fatiguing and were done at the same time of the day. They were also preceded by one familiarization session where participants became accustomed to the setup. During the first session, participants had their body mass and height measured, and their physical activity evaluated through the International Physical Activity Questionnaire-Short Form [67]; body mass was rechecked at the beginning of each experimental session. Participants were asked to maintain the same dietary pattern, avoid caffeinated beverages and food for at least 3 h before each experimental session and refrain from strenuous exercise in the 24 h before the experiment. The room temperature was kept between 23 and 25°C.

Figure 1. Experimental setup and analysis of oxygen uptake data. (a) Stair climber. For Experiment 1, participants walked on a Technogym Climb Excite stair climber at an incline of 78% and diagonal speeds of 0.20, 0.25 and 0.36 m s−1 while their gas exchanges were measured by a portable metabolic system. (b) Experimental protocol and representative oxygen uptake data. Each walking bout was preceded by a 3 min sitting rest, lasted 10–240 s and was followed by 7 min of recovery. The dashed horizontal line indicates the average resting oxygen uptake (V̇O2,rest), while the dashed vertical lines indicate the times at bout start and end. The shaded red surface indicates the area under the V̇O2 curve, that is, its definite time integral. (c) Calculation of non-metabolic phase-I oxygen transfer. The continuous red curve indicates the exponential fit of the phase-II response, while the grey area represents non-metabolic phase-I oxygen transfer (VO2pI). The dashed horizontal line indicates V̇O2,rest, while the dashed vertical line indicates the time at bout start. (d) Increase in oxygen uptake after a 30 s bout. For each bout, oxygen uptake values were smoothed through a five-breath moving average; and measure the magnitude and delay of the increase in oxygen uptake after the bout end, respectively.
(b) Experiment 2: walking on a treadmill
Ten healthy participants were recruited for a second study (five males and five females; age 28 ± 4 years, body mass 68.0 ± 12.0 kg, height 1.72 ± 0.06 m; physical activity: 1802 ± 865 MET-min week−1; six participants in common with Experiment 1). Participants walked on a motorized treadmill (Woodway, Waukesha, WI, USA) at 1.39 m s−1 for bouts lasting 10–240 s in randomized order, as for stair climbing. After sitting for 3 min, they started stepping on a treadmill whose speed was already at target. Additionally, four complete V̇O2 kinetics were collected at the same speed for bouts lasting 240 s. A familiarization session was held before the experiments.
(c) Analysis of V̇O2 kinetics
V̇O2 was averaged during the last minute of sitting before each bout to get resting oxygen uptake (V̇O2,rest) and subtracted from V̇O2 to get net oxygen uptake (V̇O2,net). EEOC and EPOC were given by the time integral of V̇O2,net from bout start to bout end and from bout end to the end of recovery, respectively (figure 1b ). The total net oxygen volume consumed for a bout (VO2(b)) was given by EEOC + EPOC, and V̇O2(b) was defined as the ratio between VO2(b) and bout duration. For stair climbing at 0.36 m s−1 and walking at 1.39 m s−1, four 240-s on-kinetics were analysed for each participant. Data from such transients were cleared from aberrant breaths (a minimum of 5 standard deviations from the local mean), interpolated (1 s time intervals), averaged [68,69] and fitted with a function in the form
where A, t d and τ are the amplitude, time delay and time constant of the monoexponential function, respectively. The first 20 s were removed to avoid fitting the cardiodynamic phase [68]. To check for the appropriateness of the monoexponential model, a biexponential one was also fitted, using the initial values and boundaries reported in electronic supplementary material, S2. The Akaike Information Criterion was calculated for each fit: the monoexponential fit was kept if a biexponential one reduced the Akaike Information Criterion by less than 10%. Oxygen volume exchanged for non-metabolic purposes during phase-I (VO2pI) was estimated as the area under the curve of V̇O2,net and above the fitted phase-II on-kinetics [43] during the first 10, 20 and 30 s (figure 1c ). Finally, the time required for V̇O2 to peak and the amplitude of such a peak were assessed: for each bout, V̇O2 was smoothed through a five-breath moving average, and if an increase in smoothed V̇O2 was observed after bout end, its magnitude () and delay from bout end () were calculated (figure 1d ).
(d) Energy consumption
The metabolic cost of transport (C met,ss; J kg−1 m−1) was calculated as [3]
where V̇O2,ss and V̇O2,rest indicate oxygen uptake during the last minute of walking and rest, respectively, EqO2,ss is the number of joules released from the steady-state oxidative combustion of 1 ml of oxygen at a given respiratory exchange ratio (RER) [70] and v is the progression speed. Equation (2.2), however, cannot be applied for short bouts that do not reach a metabolic steady state. For them, the oxygen cost of bout (C oxy(b), ml kg−1 m−1) was defined as
where b is bout duration. Historically, such an oxygen cost has been converted to a metabolic cost by assuming that EqO2 is fixed, and also that all the EEOC is used for metabolic purposes. In this case, the metabolic cost is simply a rescaled C oxy(b), and efficiency estimates are proportional to C oxy(b) −1. However, if VO2pI is not available for the reactions replenishing muscular ATP during the bout and allowing EqO2 to vary between on- and off-transients, the metabolic cost of a bout (C met(b)) can be estimated by the more general equation
For this purpose, an EqO2,on of 20.9 J ml O2 −1 and an EqO2,off of 19.6 J ml O2 −1 can be assumed to account for the different stoichiometries of ATP-producing reactions during the on- and off-transients, as in Scott [49,52]. Equation (2.4) may be modified considering that EqO2 is known after reaching a steady state: results change negligibly (electronic supplementary material, S3), and this simpler version is presented throughout the text. For stair climbing, the steady-state mechanical work to move the body centre of mass and the body segments (W mec,ss) was calculated from stereophotogrammetry [62] (electronic supplementary material, S4) as 6.4 J kg−1 m−1 at 0.36 m s−1. The same mechanical work was assumed to be 0.88 J kg−1 m−1 for treadmill walking at 1.39 m s−1, as in [71]. Hence, steady-state and bout efficiencies were defined as
To compare them with calculations from previous literature, efficiency was also alternatively calculated as
(e) Sample size estimation and statistical analysis
The sample size was estimated a priori considering the slope of the fixed effect of bout duration over V̇O2(b) as the primary endpoint and a slope equal to zero for the null hypothesis. Based on data on walking at 1.39 m s−1 from Blokland et al. [35], V̇O2(b) (ml min−1; not mass-normalized) decreases with bout duration (in seconds) as y = 870.6 − 0.37x in the range from 60 to 360 s, with a standard deviation of residuals of 22.7 ml min−1. For stair climbing bouts from 14 to 240 s, pooled data from Minetti et al. [39] and Kamon [72] show a steeper decrease, with y = 3556 − 6.1x and a standard deviation of residuals of 126.4 ml min−1. For each gait, a sample size of 6 would lead to a statistical power of 0.9 and a type-I error probability under 0.05; to account for potential data loss, the sample size was increased to 10 participants. To assess the relation of V̇O2(b), C oxy(b), C met(b) and η (b) with bout duration, such variables were regressed over bout duration with a mixed-effects model with participants as random factors. The agreement between η (b) and η ss during 240-s bouts was assessed through Bland–Altman plots [73]; to explore how steady-state duration impacted the agreement between η (b) and η ss, such analysis was also repeated for 8-min bouts in a subsample of six participants (electronic supplementary material, S3). Sample size estimation and statistical analyses were done with G*Power 3.1.9.3 [74], R 3.6.2, R Studio 1.2.5 and the package ‘lmer’ [75–77].
3. Results
(a) Oxygen uptake
One participant had a RER greater than 1 during stair climbing at 0.36 m s−1, and this condition was excluded from the analysis. Data at 0.36 m s−1 for one further participant were excluded because their on-transient was better fitted by a biexponential rather than a monoexponential model. Therefore, at 0.36 m s−1, data from eight participants were analysed. At steady state, stair climbing required an average V̇O2,net of 19.3 ± 1.5 (mean ± s.d.), 23.0 ± 1.2 and 30.1 ± 1.8 ml min−1 kg−1 for speeds of 0.20, 0.25 and 0.36 m s−1, respectively. For 240-s bouts, average V̇O2(b) was 21.1 ± 2.0, 25.3 ± 1.8 and 32.2 ± 2.8 ml min−1 kg−1 for speeds of 0.20, 0.25 and 0.36 m s−1, respectively, while it increased with decreasing bout duration at all speeds (figure 2). In the mixed-effects model, the fixed effect of bout duration over V̇O2(b) was −0.06 [−0.07; −0.04] (estimate [95% CI]; electronic supplementary material, S5). Similar observations held for treadmill walking at 1.39 m s−1, which had a V̇O2,net of 11.2 ± 1.4, V̇O2(b=240s) of 11.8 ± 1.6 ml min−1 kg−1 and a fixed effect of bout duration over V̇O2(b) of −0.07 [−0.1; −0.03] (figure 4; electronic supplementary material, S5). Non-metabolic oxygen transfer during the first 30 s was 1.4 ± 0.6 ml kg−1 for stair climbing at 0.36 m s−1 and 1.4 ± 0.3 ml kg−1 for walking at 1.39 m s−1. Further fitting results for oxygen uptake kinetics are reported in table 1. In several bouts, V̇O2 continued to increase after bout end, with a more marked effect for stair climbing bouts shorter than 60 s (table 2).

Figure 2. Mean oxygen uptake during stair climbing. For each progression speed (panels a–c), V̇O2(b) is plotted as a function of bout duration. Grey lines connect observations from the same participant. Thick red lines and black dots: between-participants averages.
variable | unit | experiment 1: stair climbing 0.36 m s−1 | experiment 2: treadmill 1.39 m s−1 |
---|---|---|---|
A | ml min−1 kg−1 | 30.1 ± 1.4 | 11.0 ± 1.0 |
td | s | 12.8 ± 66 | 8.7 ± 6.1 |
τ | s | 24.5 ± 7.6 | 26.6 ± 8.8 |
VO2pI(10s) | ml kg−1 | 0.7 ± 0.3 | 0.7 ± 0.2 |
VO2pI(20s) | ml kg−1 | 1.4 ± 0.6 | 1.3 ± 0.3 |
VO2pI(30s) | ml kg−1 | 1.4 ± 0.6 | 1.4 ± 0.3 |
C oxy(b=240s) | ml kg−1 m−1 | 1.49 ± 0.13 | 0.14 ± 0.02 |
C met(b=240s) | J kg−1 m−1 | 30.4 ± 2.5 | 2.9 ± 0.4 |
Cmet,ss | J kg−1 m−1 | 29.9 ± 1.8 | 2.8 ± 0.3 |
η (b=240s) | [dimensionless] | 0.21 ± 0.02 | 0.31 ± 0.04 |
η ss | [dimensionless] | 0.21 ± 0.01 | 0.32 ± 0.03 |
bout duration (s) | ||||||
---|---|---|---|---|---|---|
10 | 30 | 60 | 90 | 240 | ||
(ml min−1 kg−1) | Walking | |||||
1.39 m s−1 | 2.3 ± 2.0 | 1.5 ± 2.2 | 1.9 ± 1.7 | 0.5 ± 0.8 | 0.3 ± 1.0 | |
Stair climbing | ||||||
0.20 m s−1 | 4.2 ± 2.1 | 5.4 ± 3.5 | 2.0 ± 1.1 | 1.3 ± 0.8 | 0.3 ± 0.6 | |
0.25 m s−1 | 4.2 ± 2.1 | 6.5 ± 2.3 | 1.3 ± 0.8 | 0.8 ± 1.0 | 0.1 ± 0.2 | |
0.36 m s−1 | 4.3 ± 2.8 | 6.3 ± 3.1 | 1.6 ± 1.5 | 0.7 ± 1.1 | 0.0 ± 0.0 | |
(s) | Walking | |||||
1.39 m s−1 | 12 ± 11 | 6 ± 8 | 9 ± 8 | 5 ± 6 | 1 ± 2 | |
Stair climbing | ||||||
0.20 m s−1 | 35 ± 13 | 25 ± 10 | 8 ± 3 | 4 ± 2 | 1 ± 1 | |
0.25 m s−1 | 28 ± 14 | 22 ± 12 | 6 ± 4 | 5 ± 5 | 0 ± 0 | |
0.36 m s−1 | 34 ± 13 | 20 ± 10 | 6 ± 8 | 1 ± 2 | 0 ± 0 |
(b) Metabolic cost and efficiency
Oxygen cost, metabolic cost and efficiency for stair climbing at 0.36 m s−1 and treadmill walking at 1.39 m s−1 are reported in figures 3 and 4, respectively. C met,ss was 33.5 ± 2.2, 32.0 ± 1.5 and 29.9 ± 1.8 J kg−1 m−1 for stair climbing at 0.20, 0.25 and 0.36 m s−1 and 2.8 ± 0.3 J kg−1 m−1, respectively, for walking at 1.39 m s−1. η ss was 0.19 ± 0.01, 0.20 ± 0.01 and 0.21 ± 0.01 for stair climbing and 0.32 ± 0.03 for treadmill walking. For both gaits, increasing bout duration decreased C oxy (fixed effects: −2.7 × 10−3 [−4.0 × 10−3; −1.2 × 10−3] and −8.2 × 10−4 [−1.2 × 10−3; −3.9 × 10−4] for stair climbing and treadmill walking, respectively) and C met(b) (fixed effects: −3.4 × 10−2 [−6.3 × 10−2; 6.6 × 10−3] and −1.2 × 10−2 [−2.0 × 10−2; −4.2 × 10−3]) and increased η (b) (fixed effect: 1.4 × 10−4 [3.4 × 10−5; 2.4 × 10−4] and 4.7 × 10−4 [2.1 × 10−4; 7.2 × 10−4]; electronic supplementary material, S5). The longer the bouts, the more similar bout efficiency (η (b)) and steady-state efficiency (η ss): as shown by figures 5 and 6, η (b=240 s) ≈ η ss, and extant differences are further mitigated for η (b=480 s) (electronic supplementary material, S3). Figure 5 also shows that η (b) varied less than η´(b) with bout duration.

Figure 3. Oxygen cost (a), metabolic cost (b) and efficiency (c) of stair climbing at 0.36 m s−1 as a function of bout duration. Grey lines connect observations from the same participant. Thick red line and black dots: between-participants averages.

Figure 4. Oxygen uptake (a), oxygen cost (b), metabolic cost (c) and efficiency (d) for walking on a treadmill at 1.39 m s−1 as a function of bout duration. Grey lines connect observations from the same participant. Thick blue line and black dots: between-participants averages.

Figure 5. Estimates of efficiency. Steady-state efficiency (η ss; dotted lines) and bout efficiency (either η (b), solid lines or η′ (b), dashed lines) are plotted as a function of bout duration for stair climbing at 0.36 m s−1 (a) and treadmill walking at 1.39 m s−1 (b).

Figure 6. Steady-state and bout efficiency. Bland–Altman plot for bout efficiency (η (b)) versus steady-state efficiency (η ss) for 240-s bouts of stair climbing at 0.36 m s−1 (a) and treadmill walking at 1.39 m s−1 (b).
4. Discussion
In this study, we tested whether bout duration affects the oxygen and metabolic cost of human walking. Our findings indicate that shorter bouts have a substantially higher oxygen cost compared to longer ones. This greater oxygen cost may be attributed, in part, to the proportionally greater oxygen uptake for non-metabolic purposes and differences in the metabolic equivalent of oxygen during the on- and off-transients. When accounting for such factors, shorter bouts still have a lower locomotor efficiency.
(a) Bout duration and the metabolic cost of walking
The mean oxygen uptake—or V̇O2(b)—increased with decreasing bout duration for both stair climbing and treadmill walking. The same trend was observed for the oxygen cost of bout (C oxy(b)), which, at a given progression speed, is essentially a scaled V̇O2(b). These findings confirm previous observations on human and animal locomotion. Minetti et al. [39] found that people walk upstairs with a higher C oxy(b) during approximately 14 s bouts compared with approximately 25 s ones. Similarly, Blokland et al. [35] showed that C oxy(b) decreases with the duration of walking bouts lasting 60–360 s, and based on Zanconato et al. [40], C oxy(b) is greater in children and adults during 60 s bouts of walking than at steady state. The findings are also consistent with observations from comparative physiology, where similar trends have been reported for endotherms and ectotherms [9,34,36,37], even with greater duration dependency of C oxy(b) in animal studies. In this study, we confirmed such observations by investigating the impact of bout duration in humans and by standardizing bout durations and mechanical work.
To date and to the best of our knowledge, the relationship between oxygen uptake and energy expenditure for short bouts has not been elucidated: in this article, we suggest a physiological link between the two. Previous researchers relied on C oxy(b) as a proxy for metabolic cost [9,11,35,36] or converted it to the metabolic cost of bout (C met(b)) assuming a constant energy equivalent of oxygen, which simply yields a scaled C oxy(b) [39]. These approaches stem from the fact that during short bouts, the RER does not provide information about the mitochondrial respiratory quotient, and the energy equivalent of oxygen cannot be calculated. Within this bout duration and intensity range, Scott [49] theorized an average energy equivalent of 20.9 J ml O2 −1 during the on-transient and 19.6 J ml O2 −1 during the off-transient, reflecting their increased proportion of anaerobic and aerobic processes, respectively. Another challenge is considering alveolar oxygen exchanges at exercise onset. During this phase, the products of muscle metabolism have not yet reached the right heart, and mixed venous blood has not changed its composition from rest. However, the increase in cardiac output and pulmonary blood flow leads to an increase in alveolar oxygen uptake [78]. The resulting exchanged oxygen volume is not proportional to muscular energy demands [43] and was here estimated as the non-metabolic phase-I oxygen transfer (VO2pI). Consistent with the existing literature, VO2pI did not depend on exercise intensity and was similar to walking and stair climbing. Furthermore, it was exchanged almost entirely in the first 30 s, weighting proportionally more in shorter bouts. Therefore, duration-dependent differences in C met(b) are smaller than those in C oxy(b).
The remaining differences in C met(b) and bout efficiency (η (b)) between shorter and longer bouts could also be attributed to variations in the average mechanical work. Work is done to start the bout and to accelerate, but its magnitude is not enough to explain variations in metabolic cost (electronic supplementary material, S4). Additional terms are not included in the present calculations of η (b), such as ventilatory and cardiac work and ATP-dependent processes other than cross-bridge cycling, for example, active ion pumping. These processes may weigh proportionally more in shorter bouts than in longer bouts.
(b) Steady-state versus bout-based estimates of metabolic cost
Steady-state calculations of oxygen uptake and metabolic cost provide different information compared to those based on the integration method. The former calculations have a clear metabolic equivalent of oxygen consumption based on the measured RER [70] as long as exercise is below the anaerobic threshold and anaerobic contributions are negligible [27]. This is the method of choice for bouts that are long enough to reach the steady state for alveolar O2 and CO2; however, it may give inaccurate estimates of metabolic demands for shorter bouts. Indeed, for short bouts, a relevant part of the reactions that recover ATP and phosphocreatine is sustained by anaerobic processes [58], which conflates with the previous assumptions. Moreover, steady-state methods inherently assume a constant amount of oxygen volume, metabolic joules and mechanical joules required to transverse a unit distance, irrespective of the distance covered: this implies a proportionality between the oxygen volume exchanged at the alveoli, the metabolic energy expended and the mechanical work done. However, locomotion bouts also require both ATP for processes that do not directly produce mechanical work and oxygen for processes that do not proportionally replenish ATP [43,45]. In other words, extrapolations from steady-state calculations may be consistent with the ‘classic’ oxygen deficit-debt view, but not with current knowledge on the additional processes that sustain oxygen uptake during and after exercise [11,34,37].
The integration method, conversely, incorporates both steady-state oxygen costs and those additionally sustained during the on-transient and recovery [11]. The longer the bout duration, the smaller these costs’ proportional impact and the more results from the steady-state and integration methods converge (figures 5 and 6; electronic supplementary material, S3). Such a convergence confirms observations from Blokland et al. [35] for flat walking and Jessup et al. [79] for gradient walking. An alternative and non-mutually exclusive explanation for such convergence is that, within these bout durations, increasing the total locomotion time decreases the cost of locomotion independently of the proportion spent during transients. This explanation is supported by evidence from lizards [80] and mice, where increasing the frequency of multiple bouts has a similar effect to increasing the duration of continuous bouts [81]. The integral method, nonetheless, falls short in distinguishing whether costs are sustained during the bout or the recovery. Applying the integral method to the EEOC or EPOC only would have little meaning: for instance, for 30-s stair climbing bouts, EEOC was approximately 20% of VO2(b) and, if considered alone, would lead to estimating thermodynamically implausible efficiencies of approximately 1.00. Moreover, sharply distinguishing between EEOC and EPOC may be misleading: their underlying physiological processes are not fully distinct [27,45] and a delay is present between the end of the bout and the end of the oxygen on-transient (table 2). Such a delay had already been documented [39,53,55], and present data suggest that it is dependent on the duration and intensity of the bout; further work is required to elucidate its determinants.
(c) Efficiency
Bout duration had a smaller impact on efficiency than on oxygen cost (figures 3 and 4). Compared with stair climbing at 0.36 m s−1 for 240 s, doing it for 10 s bouts increased C oxy(b) by 1.6-fold and C met(b) by 1.4-fold and further decreased η (b) by 1.3-fold (figure 3). For walking at 1.39 m s−1, walking for 10 s bouts increased C oxy(b) by 3.4-fold and C met(b) by 2.9-fold and decreased η (b) by 2.2-fold (figure 4). This observation supports the view that a greater C oxy(b) does not strictly imply a proportionally greater metabolic demand and mechanical work output.
When bouts are long enough, the integral method and steady-state calculations also yield coherent efficiency estimates. For 4-min stair climbing, η (b=240 s) ≈ η ss ≈ 21%, as for uphill walking in Minetti et al. [63]: at such slopes, most metabolic work is converted into either heat or net positive mechanical work, and locomotor efficiency is similar to that of concentric muscle contraction [39,59,63]. In turn, level walking benefits from pendular energy recovery, storage and release of elastic energy, advantageous contraction speeds and gearing of propulsor muscles [82–87]: η (b=240s) ≈ η ss ≈ 32%, as in Cavagna & Kaneko and Minetti et al. [56,63]. Of note, estimates of mechanical work for level walking vary depending on underlying assumptions, including those about double support work [88,89], energy transfer between the limbs and body centre of mass [62], elastic work [84], frictional internal work [90] and soft tissue deformations [91]. Hence, the present estimate of walking efficiency should not be interpreted as the efficiency of muscular contraction during walking. For both gaits, η′ (b=240 s) < η (b=240 s) < η ss, which supports accounting for phase-I oxygen exchange and variations in the energy equivalent of oxygen when estimating bout efficiency. Neglecting them yields an estimated efficiency that is simply proportional to V̇O2(b) −1, underestimating its value and overestimating its increase with bout duration (figure 5).
However, trends in bout efficiency are only partially reconciled with those in isolated limb contractions, where efficiency may even decrease with increasing exercise duration [92]. The contradiction may appear less stark, considering that efficiency measured in isolated muscles or limbs is only one of the components contributing to whole-body locomotor efficiency. Whole-body efficiency is the product of transmissive efficiency (the ratio of muscle mechanical work to whole-body mechanical work) and muscle efficiency (the ratio of metabolic energy to mechanical work produced by muscle) [93], the latter being the product of the efficiency of converting free energy from ATP into mechanical work and that of transferring free energy from substrates to high-energy ATP bonds [94]. Isolated muscle studies assess either muscle efficiency or its components, while isolated limb studies assess the product of muscle efficiency and their task-specific transmissive efficiency. During locomotion, the efficiency of recovery reactions may vary over time [51]. However, the determinants of transmission efficiency, including elastic mechanisms [87,95], muscle coactivations, soft tissue deformations and joint frictions [66,90,96], are not known to vary within the considered time ranges. Some of them were also partially controlled for in the stair climbing experiment, where elastic and negative work are negligible and stride frequency is fixed. Factors impacting variations in efficiency in isolated studies may also weigh less in whole-body studies. For instance, variations in pH and temperature [92,97,98] are thought to impact efficiency in isolated limb and muscle studies, but their variation during locomotion bouts at such intensities may have a smaller effect on efficiency. Increased cardiac output and ventilation after exercise may also be a source of increased mechanical work [99,100], which may weigh more on shorter bouts and is not considered in the present efficiency calculations.
(d) Significance and perspectives
This article assessed the energy demands of short bouts of walking and stair climbing: the same framework can be applied when a steady state is not reached owing to changes in speed, direction and slope [79,101] or when physical conditions slow down its achievement [42]. Moreover, the increasing recognition of short locomotor bouts as a means to promote physical activity [16,21] highlights the need to better quantify their intensity. Based on the present results, steady-state measures such as the MET [102] may underestimate the demands of short walking bouts, for which new metrics may be needed. The greater metabolic demands of short bouts could also explain the impact of interventions based on intermittent locomotion and ‘exercise snacks’.
Finally, this work sheds light on the analogies and differences in the ‘move now, pay later’ strategy of humans versus other animals. Such a mechanism has greater importance for ectothermic animals with a small aerobic scope [103]. However, the shorter the bouts, the more intermittent the locomotion pattern and the lower the aerobic fitness of people, the greater its relevance for human walking.
(e) Methodological considerations
V̇O2 was calculated using a Cosmed K5 device, whose algorithm neglects within-breath changes in alveolar oxygen stores. By correcting them, alternative methods may reduce variability in breath-by-breath estimates [104–106]. However, averaging consecutive breaths, as during integration, makes such correction unnecessary since net changes in lung stores are negligible compared to the exchanged volume of O2 [104,107]. Here, the main focus was total exchanged volume rather than the shape of oxygen kinetics; hence, a classical breath-by-breath approach was preferred. Such an approach allowed using a portable metabolic system and avoiding potential gait changes introduced by using a fixed one.
This study standardized mechanical work instead of exercise intensities: for the same stair climbing speed, intensity domains may differ between participants, with potentially different oxygen responses and proportion of anaerobic contributions [108]. Standardizing mechanical work is necessary in locomotion studies to link metabolic costs to progression speed and allow meaningful efficiency calculations; moreover, exercise thresholds are activity-dependent [109] and could be hard to measure in a reliable way for walking and stair climbing. However, the potential impact of exercise intensity was indirectly considered in two ways. First, only acquisitions with a RER lower than one were included in the analysis. Moreover, the major findings for V̇O2(b), C oxy(b), C met(b), η (b) and VO2pI were confirmed in a range of activities where V̇O2,ss spanned approximately 10–30 ml min−1 kg−1.
Finally, EPOC estimates of metabolic cost could be corrected based on blood lactate concentration [110]. While acknowledging the theoretical rationale for this ‘lactate method’ here, we opted not to implement it for several reasons. First, muscle lactate concentrations differ from blood lactate measurements [111,112], hindering their interpretability in this study. The ‘lactate method’ also requires distinguishing alactic and lactic components of V̇O2 off-kinetics [101], with poorly reliable results for short bouts with relatively low intensity. Moreover, measuring lactate could interfere with breathing patterns after the bout; such interference appears more justified the higher the bout intensities, the longer the bout durations and the higher the RER during corresponding steady-state exercise. However, for short stair climbing bouts, Minetti et al. [39] reported negligible elevation in blood lactate, and Scott [49] showed that within the same range of intensities and durations, the ‘lactate method’ and the ‘variable energy equivalent of oxygen method’ yield equivalent results.
5. Conclusion
Shorter walking bouts incur a substantially greater oxygen cost than longer ones. Such a discrepancy may be owing to their greater proportion of oxygen uptake for non-metabolic purposes, variations in the energy equivalent of oxygen and lower efficiency. Short walking bouts are the most frequent form of locomotion for humans and many animals, and this study sheds light on the energy requirement of such brief activities. By doing so, it also offers a framework for assessing the metabolic cost of activities that do not reach a metabolic steady state.
Ethics
The investigation was done with the permission of the Institutional Review Board of the University of Milan (03.15.11).
Data accessibility
Supplementary data contain individual participant data used for the images and tables in the text.
Supplementary material is available online [113].
Declaration of AI use
Yes, we have used AI-assisted technologies in creating this article.
AI-assisted technologies were used to proofread English spelling and phrasing and as a search engine.
Authors’ contributions
F.L.: conceptualization, data curation, formal analysis, investigation, methodology, software, validation, visualization, writing—original draft, writing—review and editing; L.R.: conceptualization, investigation, writing—review and editing; A.E.M.: conceptualization, funding acquisition, project administration, supervision, writing—review and editing; G.P.: conceptualization, funding acquisition, investigation, methodology, project administration, software, supervision, validation, writing—original draft, writing—review and editing.
All authors gave final approval for publication and agreed to be held accountable for the work performed therein.
Conflict of interest declaration
We declare we have no competing interests.
Funding
No funding has been received for this article.
Acknowledgements
We thank Dr Silvano Zanuso, head of R&D department of Technogym, for lending us the stair climber. We also thank Elena Lissoni and Gabriele Lucchi for helping in data collection.