Cycling efficiency and energy cost of walking in young and older adults

Abstract

To determine whether age affects cycling efficiency and the energy cost of walking (Cw), 190 healthy adults, ages 18–81 yr, cycled on an ergometer at 50 W and walked on a treadmill at 1.34 m/s. Ventilation and gas exchange at rest and during exercise were used to calculate net Cw and net efficiency of cycling. Compared with the 18–40 yr age group (2.17 ± 0.33 J·kg⁻¹·m⁻¹), net Cw was not different in the 60–64 yr (2.20 ± 0.40 J·kg⁻¹·m⁻¹) and 65–69 yr (2.20 ± 0.28 J·kg⁻¹·m⁻¹) age groups, but was significantly (P < 0.03) higher in the ≥70 yr (2.37 ± 0.33 J·kg⁻¹·m⁻¹) age group. For subjects >60 yr, net Cw was significantly correlated with age (R² = 0.123; P = 0.002). Cycling net efficiency was not different between 18–40 yr (23.5 ± 2.9%), 60–64 yr (24.5 ± 3.6%), 65–69 yr (23.3 ± 3.6%) and ≥70 yr (24.7 ± 2.7%) age groups. Repeat tests on a subset of subjects (walking, n = 43; cycling, n = 37) demonstrated high test-retest reliability [intraclass correlation coefficients (ICC), 0.74–0.86] for all energy outcome measures except cycling net energy expenditure (ICC = 0.54) and net efficiency (ICC = 0.50). Coefficients of variation for all variables ranged from 3.1 to 7.7%. Considerable individual variation in Cw and efficiency was evident, with a ~2-fold difference between the least and most economical/efficient subjects. We conclude that, between 18 and 81 yr, net Cw was only higher for ages ≥70 yr, and that cycling net efficiency was not different across age groups.

NEW & NOTEWORTHY This study illustrates that the higher energy cost of walking in older adults is only evident for ages ≥70 yr. For older adults ages 60–69 yr, the energy cost of walking is similar to that of young adults. Cycling efficiency, by contrast, is not different across age groups. Considerable individual variation (∼2-fold) in cycling efficiency and energy cost of walking is observed in young and older adults.

INTRODUCTION

The effect of age on the efficiency of muscular work has been a topic of debate (22, 34). It has consistently been reported that the energy cost of walking (Cw) is ~10–30% higher in older adults (11, 13, 17–20, 23–25, 30, 36). The increased Cw in older adults is thought to be due in part to energy costs associated with greater antagonist muscle coactivation in older adults (11, 20, 23–25), secondary to reduced coordination in this population (12). By contrast, stationary cycling is a less complex mode of exercise, and is not influenced by balance and gait issues associated with walking. Thus cycle ergometer exercise may be preferred to walking for purposes of determining whether muscular efficiency is affected by age. Unlike the consistent results for treadmill walking, the effect of age on cycling efficiency is unclear, with studies reporting lower (3, 6, 10), equal (2, 21), and higher (35) efficiency in older adults.

Another issue that has not been rigorously examined is whether the age range of older adults influences the comparison of energy cost of exercise in young and older adults. Although Cw has consistently been reported to be higher in adult populations with mean ages >70 yr (11, 13, 17, 18, 20, 23–25), whether Cw is higher in older adults <70 yr is less certain (17, 19, 30). One large-scale study of adults ages ≥40 yr reported that higher gross Cw (resting metabolic rate was not measured) at a self-selected pace on an indoor 20-m course was evident only after age 65 yr (30). In two, more well-controlled studies that examined young and older adults at five different treadmill walking speeds, subjects with mean (± SD) ages of 65.3 ± 2.5 yr (17) and 67.1 ± 4.6 yr (19) generally had higher net Cw compared with younger adults. However, in one of these studies (17) the net Cw was only statistically significantly higher in older subjects at two of the five treadmill speeds studied. For cycling, we are unaware of any comparisons of efficiency across older age groups stratified by age.

Thus the purpose of this study was to examine whether Cw and cycling efficiency differed across age groups in a large sample of healthy adults, ages 18–81 yr. We hypothesized that Cw would be higher in older adults (60–81 yr) compared with younger adults (18–40 yr), and that cycling efficiency would be unrelated to age.

METHODS

Subjects.

These data were collected as part of a National Institutes of Health (NIH)-funded project that required measurement of energy expenditure (EE) in adults engaged in various physical activities (R01-HL-091006). A total of 232 healthy nonsmoking adults were recruited. Subjects were excluded if they smoked or failed to meet requirements of the Physical Activity Readiness Questionnaire (PAR-Q) (32). If subjects answered “yes” to any of the questions on the PAR-Q, they were required to obtain a physician’s clearance to participate in the study. Due to recruitment objectives for the physical activity energy expenditure project, >90% of the volunteers were between the ages of 18–40 yr and ≥60 yr. In an effort to achieve a distinct difference in age between young and older subjects, we excluded data on 18 subjects ages 41–59 yr. Further exclusion criteria for final analyses included missing/incomplete data, equipment malfunction, and any subject who exhibited an oxygen uptake (V̇o₂) slow component during walking or cycling (Fig. 1). Subject characteristics for walking (n = 190) and cycling (n = 168) experiments are displayed in Tables 1 and 2. This study was approved by the Arizona State University Institutional Review Board and conformed to the ethical standards of the Declaration of Helsinki. All subjects provided written informed consent before participation.

Fig. 1.

Flow of subjects and number excluded before final data analysis.

Table 1.

Subject characteristics: treadmill walking

			Older Adults
	All Subjects	Young Adults (18–40 yr)	60–64 yr	65–69 yr	≥70 yr	P Value
n	190	96	31	34	29
Sex, male/female	86/104	39/57	12/19	17/17	18/11
Age, yr	46 ± 21	26 ± 5	62 ± 1	66 ± 1	73 ± 3
Height, cm	169.4 ± 9.4	169.6 ± 10.0	166.9 ± 9.3	169.8 ± 7.9	170.8 ± 8.8	0.39
Weight, kg	73.1 ± 16.4	69.4 ± 15.0	71.5 ± 15.4	79.7 ± 17.7*	78.9 ± 17.1*	0.002
Body mass index, kg/m2	25.3 ± 4.6	24.0 ± 4.1	25.4 ± 3.1	27.5 ± 5.4*	27.0 ± 5.3*	<0.001
Body fat percent, %	27.8 ± 10.5	23.9 ± 10.1	31.5 ± 6.6**	32.1 ± 11.1*	31.8 ± 10.0*	<0.001
Fat mass, kg	20.9 ± 10.9	16.9 ± 9.6	22.3 ± 5.9*	26.4 ± 12.6*	25.9 ± 11.9*	<0.001
Fat-free mass, kg	52.2 ± 11.7	52.5 ± 12.0	49.2 ± 12.8	53.3 ± 11.2	53.0 ± 10.1	0.48

Data are presented as means ± SD P value reflects differences between groups using 1-way ANOVA. Bonferroni (variances homogeneous) or Games-Howell (variances not homogeneous) post hoc tests used.

^*Significant difference vs. 18–40 yr, P < 0.05.

^**Significant difference vs. 18–40 yr, P < 0.001.

Table 2.

Subject characteristics: cycling

			Older Adults
	All Subjects	Young Adults (18–40 yr)	60–64 yr	65–69 yr	≥70 yr	P Value
n	168	86	29	28	25
Sex, male/female	83/85	37/49	12/17	16/12	18/7
Age, yr	46 ± 21	26 ± 5	62 ± 1	66 ± 1	73 ± 3
Height, cm	170.1 ± 9.4	169.7 ± 10.3	167.8 ± 8.9	171.4 ± 7.7	173.2 ± 7.8	0.16
Weight, kg	74.5 ± 16.9	69.9 ± 15.8	72.2 ± 15.8	83.6 ± 16.1*	82.6 ± 16.1*	<0.001
Body mass index, kg/m2	25.6 ± 4.8	24.2 ± 4.4	25.4 ± 3.4	28.4 ± 5.0**	27.5 ± 5.3*	<0.001
Body fat percent, %	27.9 ± 10.4	24.0 ± 10.1	30.7 ± 6.3*	33.1 ± 10.2**	32.2 ± 10.6*	<0.001
Fat mass, kg	21.4 ± 11.2	17.2 ± 10.1	22.0 ± 6.2*	28.2 ± 12.0**	27.5 ± 12.4*	<0.001
Fat-free mass, kg	53.1 ± 11.6	52.7 ± 12.0	50.2 ± 12.7	55.4 ± 10.7	55.2 ± 9.4	0.29

Data are presented as means ± SD P value reflects differences between groups using 1-way ANOVA. Bonferroni (variances homogeneous) or Games-Howell (variances not homogeneous) post hoc tests used.

^*Significant difference vs. 18–40 yr, P < 0.05.

^**Significant difference vs. 18–40 yr, P < 0.001.

Anthropometrics.

Standing height (cm) was measured to within 0.1 cm against a wall-mounted stadiometer (Seca, Hamburg, Germany). Weight (kg) and body composition (body fat percentage, fat mass, and fat-free mass) were measured using a calibrated air displacement plethysmograph (BODPOD, Cosmed, Rome, Italy).

Experimental protocols.

Subjects were instructed to consume only water during the 3 h before each visit. As part of the protocol for the larger NIH-sponsored study, all subjects performed a 90-min physical activity routine in a climate-controlled laboratory [see Bhammar et al. (4) for details]. This routine included a variety of mostly light-intensity and moderate-intensity activities, two of which were walking on a treadmill (Trackmaster TMX 425, Full Vision, Newton, KS) at 1.34 m/s (3.0 mph) and cycling on a cycle ergometer (Viasprint 150P, Ergoline, Bitz, Germany) at 50 W. Each activity was performed for 8 min, with 4 min of rest between each activity. Subjects were not permitted to hold the handrails during treadmill walking, and cadence (60–65 rpm) was closely monitored during cycling. Prior to the physical activity routine, subjects rested in a seated position for 5 min and then stood upright for 5 min.

Each subject was fitted with a lightweight, portable metabolic measurement system (Oxycon Mobile, Becton, Dickinson, Franklin Lakes, NJ) that has been validated against the Douglas Bag Method (29). Calibration was performed before each testing visit according to manufacturer’s specifications. Pulmonary ventilation and gas exchange were measured breath-by-breath for determination of V̇o₂ and carbon dioxide production (V̇co₂), which were used to compute EE (kcal/min) (16). V̇o₂ was expressed as 30-s averages of breath-by-breath data.

A subset of the larger sample (n = 43 for walking, n = 37 for cycling) performed the same physical activity routines on two separate occasions at the same time of day to obtain test-retest reliability.

Energetic assessments during walking and cycling.

Ventilation and gas exchange during the last 5 min of each 8-min exercise bout were used for determination of V̇o₂ and EE. For resting metabolic rate we used the final 2 min of each 5-min measurement period for seated and standing rest.

Gross metabolic rate (W/kg) during walking at 1.34 m/s and during cycling at 50 W were calculated from V̇o₂ and V̇co₂ using the associated nonprotein respiratory quotient (7, 16). Net V̇o₂ (ml·kg⁻¹·min⁻¹), net EE (kcal·kg⁻¹·min⁻¹), and net metabolic rate (W/kg) during walking at 1.34 m/s were calculated by subtracting corresponding standing resting values from each of the gross values during walking. Gross and net energy cost per meter (J·kg⁻¹·m⁻¹) were calculated by dividing gross and net metabolic rate (W/kg) by walking speed (1.34 m/s).

Net V̇o₂ (ml/min) and net EE (kcal/min) during cycling at 50 W were calculated by subtracting corresponding seated resting values from the gross values during cycling. Gross and net efficiency (%) during cycling were calculated according to Gaesser and Brooks (7).

Statistical analysis.

All statistical analyses were conducted using SPSS 21 (IBM, Armonk, NY) with significance set at P < 0.05. Descriptive statistics are presented as means ± SD. To ensure that only steady-state V̇o₂ was used to calculate EE, we eliminated from analysis any subject’s data for which we could confirm a slow component of V̇o₂ (8). To identify a V̇o₂ slow component, we performed linear regression of V̇o₂ over time during the final 5 min of exercise for all exercise bouts. If the regression beta coefficient for V̇o₂ over time revealed a positive (β > 0) and significant (P < 0.05) slope, that session was excluded from data analysis. Sample distribution normality was assessed with the Kolmogorov-Smirnov test (K-S Test). If distribution was not normal, variables were log-transformed and K-S test repeated to ensure normality before analysis.

Based on previous reports (17, 30), we stratified subjects ≥60 yr into three age groups (60–64 yr, 65–69 yr, ≥70 yr). Levene’s Homogeneity Test was conducted to assess homogeneity of variance across age groups for each variable of interest. Analysis of variance with Bonferroni (equal variance across groups) or Game-Howell (unequal variances across groups) post hoc tests were performed to assess whether there were significant (P < 0.05) differences between the four age groups for Cw and cycling efficiency. Regression analysis was used to examine whether age was associated with Cw and cycling efficiency within young and older age groups.

Test-retest reliability for walking (n = 43) and cycling (n = 37) measures were assessed by two-way, random-effects, single-measure intraclass correlation (ICC) between visit 1 and visit 2. The ICC is a better indicator of stability over time than Pearson product-moment correlation as it assesses both variance explained by individuals and mean differences over time (9). In addition, coefficients of variation (CVs) were assessed for walking and cycling measures between visits.

RESULTS

Data from 190 subjects (96 young; 94 older) were used for comparing Cw (Table 1), and data from 168 subjects (86 young, 82 older) were used for comparing cycle ergometer efficiency (Table 2). The higher body weights of subjects ≥ 65 yr were due to greater fat mass. Fat-free mass was not different across all age groups.

For walking, net EE and net Cw were significantly higher (both P < 0.03) in the ≥70 yr age group compared with the 18–40 yr age group (Table 3). Net EE and net Cw did not differ between the three <70 yr age groups (Table 3). For subjects ≥60 yr, net Cw was significantly related to age (R² = 0.123, P = 0.002), with the relationship between net Cw and age best fit by a quadratic equation (Fig. 2). For younger adults, net Cw was unrelated to age (R² = 0.023, P = 0.137) (Fig. 2).

Table 3.

Energy cost of walking at 1.34 m/s and cycling efficiency at 50 W in young (18–40 yr) and older (60–81 yr) adults

		Older Adults
	Young Adults (18–40 yr)	60–64 yr	65–69 yr	≥70 yr	P Value
Walking economy (n = 190)	n = 96	n = 31	n = 34	n = 29
Gross V̇o2, ml·kg−1·min−1	12.7 ± 1.7	12.3 ± 1.8	12.2 ± 1.4	12.6 ± 1.3	0.33
Rest V̇o2, ml·kg−1·min−1	4.1 ± 0.8	3.5 ± 0.6*	3.5 ± 0.7**	3.3 ± 0.7**	<0.001
Net V̇o2, ml·kg−1·min−1	8.6 ± 1.4	8.8 ± 1.6	8.8 ± 1.2	9.3 ± 1.3	0.11
Gross EE, kcal·kg−1·min−1	0.062 ± 0.007	0.059 ± 0.009	0.059 ± 0.006	0.062 ± 0.006	0.21
Net EE, kcal·kg−1·min−1	0.042 ± 0.006	0.042 ± 0.008	0.042 ± 0.005	0.046 ± 0.006*	0.04
Gross energy cost/distance, J·kg−1·m−1	3.22 ± 0.42	3.09 ± 0.47	3.09 ± 0.33	3.21 ± 0.34	0.21
Net energy cost/distance, J·kg−1·m−1	2.17 ± 0.33	2.20 ± 0.40	2.20 ± 0.28	2.37 ± 0.33*	0.04
Cycling efficiency (n = 168)	n = 86	n = 29	n = 28	n = 25
Gross V̇o2, ml/min	910 ± 120	850 ± 150	920 ± 120	840 ± 90	0.01
Rest V̇o2, ml/min	290 ± 80	250 ± 70	280 ± 80	260 ± 60	0.05
Net V̇o2, ml/min	620 ± 80	600 ± 110	630 ± 90	580 ± 70	0.10
Gross EE, kcal/min	4.52 ± 0.55	4.22 ± 0.69	4.55 ± 0.58	4.24 ± 0.46	0.02
Net EE, kcal/min	3.11 ± 0.39	3.01 ± 0.51	3.16 ± 0.45	2.95 ± 0.33	0.19
Gross efficiency, %	16.1 ± 2.0	17.4 ± 2.5*	16.0 ± 2.2	17.1 ± 1.8	0.01
Net efficiency, %	23.5 ± 2.9	24.5 ± 3.6	23.3 ± 3.6	24.7 ± 2.7	0.18

Data are presented as means ± SD. EE, energy expenditure; V̇o₂, oxygen uptake. P value reflects differences between groups using 1-way ANOVA. Bonferroni post hoc tests used.

^*Significant difference vs. 18–40 yr, P < 0.05.

^**Significant difference vs. 18–40 yr, P < 0.001.

Fig. 2.

Individual data for net energy cost per distance (J·kg⁻¹·m⁻¹) walking at 1.34 m/s in young (18–40 yr) and older (60–81 yr) adults.

For cycling, net EE and net efficiency did not differ between groups (Table 3). Gross efficiency was significantly higher in the 60–64 yr age group compared with the 18–40 yr age group (P = 0.03). Net efficiency was positively correlated with age in the 18–40 yr group (R² = 0.047, P = 0.045), but was not significantly related age in older subjects (R² = 0.041. P = 0.189) (Fig. 3).

Fig. 3.

Individual data for net efficiency (%) cycling at 50 W in young (18–40 yr) and older (60–81 yr) adults.

For both age groups there was considerable individual variation in both Cw and cycling efficiency, with distributions that were similar for both age groups (Figs. 2 and 3). Aside from one outlier in each of the young and older age groups, during walking the least economical young and older subjects had net energy costs of walking (~3 J·kg⁻¹·m⁻¹) that were approximately twice that of the most economical subjects (~1.5 J·kg⁻¹·m⁻¹) (Fig. 2). Similarly for cycling, the most efficient subjects had net efficiencies (~30%) that were approximately twice that of the least efficient subjects (~15–18%) (Fig. 3).

For test-retest reliability, 37 subjects (18 young, 19 older) performed two identical exercise sessions cycling at 50 W, and 43 subjects (23 young, 20 older) performed two identical treadmill walking sessions at 1.34 m/s (Table 4). For all walking gross and net EE and Cw measures, ICCs ranged between 0.80 and 0.86. For resting EE while standing, the ICC was 0.74. For cycling, ICCs were between 0.83 and 0.85 for seated resting EE and gross EE and gross efficiency. The ICCs for net EE (0.54) and net efficiency (0.50) were lower. Coefficients of variation for all walking and cycling measures ranged between 3.1% and 7.7% (Table 4).

Table 4.

Test-retest reliability for walking at 1.34 m/s (n = 43) and cycling at 50 W (n = 37) in a subset of adults

	Visit 1	Visit 2	ICC	CV, %
Walking at 1.34 m/s (n = 43)
Rest EE, kcal·kg−1·min−1	0.019 ± 0.004	0.019 ± 0.004	0.74	7.7 ± 5.8
Gross EE, kcal·kg−1·min−1	0.061 ± 0.007	0.060 ± 0.007	0.86	3.3 ± 2.3
Net EE, kcal·kg−1·min−1	0.042 ± 0.006	0.041 ± 0.007	0.80	5.6 ± 3.5
Gross metabolic cost, J·kg−1·m−1	3.15 ± 0.35	3.13 ± 0.35	0.86	3.4 ± 2.3
Net metabolic cost, J·kg−1·m−1	2.17 ± 0.32	2.15 ± 0.34	0.80	5.7 ± 3.6
Cycling at 50 W (n = 37)
Rest EE, kcal/min	1.35 ± 0.37	1.35 ± 0.31	0.85	7.7 ± 6.8
Gross EE, kcal/min	4.37 ± 0.42	4.34 ± 0.43	0.84	3.1 ± 2.5
Net EE, kcal/min	3.02 ± 0.32	2.9 ± 0.35	0.54	6.5 ± 3.9
Gross efficiency, %	16.6 ± 1.7	16.6 ± 1.8	0.83	3.3 ± 2.6
Net efficiency, %	24.1 ± 2.7	24.0 ± 2.9	0.50	6.8 ± 4.1

Data are presented as means ± SD. Test-retest reliability ICCs assessed by 2-way random effects single-measures ANOVA.

DISCUSSION

We hypothesized that net Cw would be higher in older adults whereas net efficiency for cycling would be similar across age groups. Our results confirmed our hypothesis regarding cycling efficiency. With regard to the energy cost of walking, our results did not completely confirm our hypothesis since higher net Cw was only evident in older adults ≥70 yr. For adults 60–69 yr, net Cw was not different from that of younger adults.

Effects of age on energy cost of walking.

Previous studies of the effects of age on the energy cost of walking have consistently reported higher net Cw in older adults (11, 13, 17–20, 23–25, 30, 36). These studies generally show that net Cw is 10–30% higher in older adults. It is important, however, to distinguish the age range of study participants. In studies comparing young and older subjects, most have used just one older age group, typically with mean ages between 65 and 82 yr (11, 13, 17–20, 23–25, 30, 36). Malatesta et al. (17) reported that net Cw in older adults with a mean age of 81.6 ± 3.3 yr was 22% higher than that of young adults (mean age 24.6 ± 2.6 yr), but that net Cw in older adults with a mean age 65.3 ± 2.5 yr was only higher (~12%) than the net Cw of the young adults in two of five walking speeds studied. Consistent with this, Schrack et al. (30) recently reported that the higher Cw in older adults is highly dependent on the ages of older adults. In this cross-sectional study, between 65 and 70 yr gross Cw was only slightly higher than that of younger adults ages 40 to <65 yr. Beyond 70 yr, gross Cw increased nonlinearly within the age range of 70 yr to the highest age of ~95 yr. However, Schrack et al. (30) did not measure resting metabolic rate, so net Cw was not assessed. Moreover, the test consisted of walking at a self-selected pace around a 20-m indoor course for only 2.5 min, with V̇o₂ during the final minute used for Cw determination. Thus it cannot be certain that steady-state V̇o₂ was achieved. Also, since Cw is influenced by walking speed (17, 18, 20, 24, 25), using self-selected walking speed complicates interpretation of the findings of Schrack et al. (30).

When assessing Cw beyond age 70 yr, our results on net Cw are consistent with previous studies (11, 13, 17–20, 23–25, 30, 36). Published data on net Cw for older adults <70 yr are limited (17, 19), with both of these studies reporting higher net Cw in adults with mean (±SD) ages 65.3 ± 2.5 yr (17) and 67.1 ± 4.6 yr (19). It is not certain that all older subjects in these two populations, particularly that of McCann et al. (19), were <70 yr, which could have influenced their results. Our data indicate that net Cw is independent of age up to 70 yr.

As our study was designed to only assess EE in adults while performing physical activity, we did not perform any biomechanical assessments that might help explain the higher Cw in older adults. Although there is no definitive explanation for the higher Cw in older adults, it has been suggested that greater coactivation of antagonist muscles may be responsible for the greater Cw in older adults (11, 20, 23, 25). The higher Cw in older adults is apparently not due to the amount of mechanical work performed by the limbs. In fact, older adults may actually perform less individual limb work to lift and accelerate body center of mass compared with young adults, but expend more energy to do so (24).

Effects of age on cycling efficiency.

Published data on the effects of age on cycling efficiency are sparse by comparison to the effects of age on the energy cost of walking, and are not as consistent. Older subjects have been reported to have lower (3, 6, 10), equal (2, 21), and higher (35) cycling efficiency compared with younger subjects. In two of the studies that reported lower cycling efficiency in older adults (3, 10), the conclusions are invalid due to methodological flaws, as described below. The lower reported cycling efficiency in older adults in these two studies was based on EE during cycling at the same relative power outputs. Since the older adults had lower maximum aerobic capacities, they cycled at lower absolute power outputs. It is well documented that gross and net efficiency increase as a function of power output, due to the fact that resting EE becomes a smaller proportion of the total exercise EE at higher power outputs (7). Thus the lower reported cycling efficiencies of older adults (3, 10) are artifacts resulting from exercise at lower absolute power outputs. For example, although older subjects had lower gross efficiency cycling at 50% and 60% of maximum aerobic power, gross efficiency was not different when comparing older and younger adults during cycling at 100 W, regardless of the training status of the subjects (10). Additionally, at 150 W gross efficiency of untrained younger subjects was 17.9% compared with 18.8% for untrained older subjects (10). These results are incompatible with the notion that cycling efficiency is lower in older adults.

Delta efficiency, which can be calculated from the change in EE associated with a change in power output (7), or the inverse of the slope of V̇o₂ on power output during incremental exercise (26), provides a better estimate of muscular efficiency. Two studies have reported that delta efficiency, defined as the inverse of the V̇o₂-power slope, is not different when comparing young and older adults (2, 21). By contrast, Conley et al. (6) reported an 18.5% lower delta efficiency in older adults compared with younger adults during a cycle ergometer ramp test. In female centenarians, Venturelli et al. (35) reported a significantly lower V̇o₂-power slope during incremental exercise. Delta efficiency was not calculated in that study, but data from their incremental tests indicate a V̇o₂-power slope of ~3.5 ml·min⁻¹·W⁻¹, which would yield a delta efficiency of >80% (see figure 5 in Ref. 35). This slope is only about one-third of the typical range of 10–12 ml·min⁻¹·W⁻¹ (7, 21, 26), and the delta efficiency nearly three times that reported for young adults (7, 21, 26) and older adults with a mean age of 71.4 yr (2). It is possible that the extremely low V̇o₂-power slope reported in female centenarians was due in part to very slow V̇o₂ kinetics, as an age-related decline in V̇o₂ kinetics is well established (27). Moreover, Venturelli et al. (35) used a 1-min incremental exercise test for the centenarians. The relatively short work rate intervals, coupled with slower V̇o₂ kinetics, increases the likelihood of a substantially reduced V̇o₂-power slope in the centenarians. Resolution of this issue would require that subjects perform a series of constant-power tests of sufficient duration to ensure attainment of steady-state V̇o₂.

Our finding of similar net efficiency in young and older adults would not be predicted on the basis of results from several studies of mitochondrial function during exercise in older adults (6, 14, 15). Reduced mitochondrial function in skeletal muscle of older adults has been documented (1). In comparing older (mean ages 71–73 yr) and younger (mean ages 22–25 yr) subjects during dynamic knee extension (14) and plantar flexion (15) exercise, older adults had two to three times higher ATP cost per contraction. Moreover, Conley et al. (6) used ³¹P-magnetic resonance spectroscopy during electrical stimulation of quadriceps muscle to show that a decline in mitochondrial coupling efficiency (with no difference in contraction-coupling efficiency) was the most likely explanation for the reduced delta efficiency they observed during a cycle ergometer ramp exercise protocol in older adults. The reduced mitochondrial coupling efficiency was associated with a lower phosphorylation capacity of mitochondria in older adults, which implies that impaired mitochondrial function may contribute to reduced muscular efficiency in older adults while exercising.

Our results, and those of others (2, 10, 21, 35), indicate that efficiency, as assessed by indirect calorimetry during cycle ergometry, is not impaired in older adults. If ATP cost of muscle contractions is elevated in older adults (6, 14, 15), there must be compensating mechanisms to account for similar, or higher, cycling efficiencies observed in older adults. Net EE and net efficiency were not different across our four age groups. In fact, gross efficiency was higher in the 60–64 yr age group compared with younger adults (Table 3). The positive correlation between age and net efficiency in the young adults was unexpected. We hesitate to ascribe any significance to this correlation as <5% of the variance in efficiency was explained by age (Fig. 3).

Reliability of measurements.

The ICCs for net and gross EE and Cw during the test-retest walking trials (all ≥ 0.80) indicate excellent reliability. This is important to demonstrate because it has been previously reported that metabolic rate was reduced in a second walking trial in young and older adults (18, 33) even when subjects were given a 30-min familiarization session on the treadmill before the first walking trial during which EE was measured (18). For cycling, ICCs for seated resting EE, gross EE, and gross efficiency were also excellent (all ≥ 0.83). Thus it was surprising that ICCs for net EE (0.54) and net efficiency (0.50) were not as strong. Nonetheless, the results for net EE and net efficiency generally matched those for gross EE and gross efficiency (Table 2), suggesting that our net EE and net efficiency results are robust.

The excellent reliability for walking energy expenditure also suggests that the considerable individual variation in net Cw (Fig. 2) is reproducible. Although the ICC for net efficiency of cycling was only fair, the high ICCs for resting and gross EE and gross efficiency also suggest that the marked individual variation in cycling efficiency (Fig. 3) is reproducible. The substantial, and similar, ranges of Cw and cycling efficiency evident in both young and older adults illustrates the challenges associated with estimating energy cost of physical activity on an individual level.

Strengths and limitations.

Our large sample size is a strength, which allowed us to examine differences not only between young and older adults, but also to determine whether age-related differences existed within a relatively large number of adults > 60 yr. Having subjects perform both walking (weight-bearing) and cycling (non-weight-bearing) exercise is also a strength, as this has not been done in previous studies. Individual data were rigorously screened to eliminate any tests that exhibited a V̇o₂ slow component, thereby ensuring steady-state EE. Our findings are also strengthened by generally high reliability provided by our test-retest results.

Our study included only one walking speed and one cycling power output, which precluded calculation of delta efficiency. The use of net efficiency, however, provides a suitable means of comparing the net energy cost of performing an identical cycling task in young and older adults. The walking speed utilized (1.34 m/s) is very close to preferred walking speed of adults and is the same as (17, 18, 25) or similar to (19, 20, 23, 24), that used in other studies. As our only dietary control, we asked subjects to consume only water during the 3 h before each visit. This may not have been sufficient to completely eliminate any lingering effect of dietary-induced thermogenesis produced by food or energy-containing beverages that subjects may have consumed just before that 3-h period (28, 31). However, our subjects did not begin the exercise protocol until ~1 h after arrival to the laboratory, and dietary-induced thermogenesis is largely over by ~4 h postprandial (28, 31). Consumption of caffeine-containing beverages before the 3-h water-only period could possibly have affected resting energy expenditure, but caffeine has been reported to have no effect on exercise energy expenditure (5). The large number of subjects in our study, along with our test-retest reliability data, reduce the likelihood that our results are confounded by not including more rigorous dietary controls. Finally, we included only two subjects >80 yr, which limits comparison to data published on centenarians (35).

Conclusions.

We conclude that, within the age range of 18–81 yr, net Cw is only higher for ages ≥70 yr. Unlike results from previous studies (17, 19, 30), our data demonstrate that for older adults ages 60–69 yr, net Cw is similar to that of young adults. For cycling, net efficiency does not differ between young and older adults.

GRANTS

This work was supported by National Heart, Lung, and Blood Institute Grant R01-HL-091006.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.A.G. conceived and designed research; W.J.T., B.J.S., D.M.B., and S.S.A. performed experiments; W.J.T. and S.S.A analyzed data; G.A.G., W.J.T., B.J.S., D.M.B., and S.S.A. interpreted results of experiments; W.J.T. prepared figures; G.A.G. and W.J.T. drafted the manuscript; G.A.G., W.J.T., B.J.S., D.M.B., and S.S.A. edited and revised manuscript; G.A.G., W.J.T., B.J.S., D.M.B., and S.S.A. approved final version of manuscript.