ARTE index revisited: linking biomarkers of cardiometabolic health with free-living physical activity in postmenopausal women

Abstract

Activity-related energy expenditure (AEE) correlates with physical activity volume; however, between-person differences in body size and walking economy (net V̇o₂) can influence AEE. The ratio of total energy expenditure (TEE) and resting energy expenditure (REE) estimates physical activity level (PAL) relative to body mass, yet does not account for variance in walking economy. The activity-related time equivalent (ARTE_walk) circumvents such constraints by adjusting for individual-specific walking economy. Herein, we compared AEE, PAL, and ARTE_walk index in a cohort (n = 81) of postmenopausal women while examining possible associations with biomarkers of cardiometabolic health. Secondary analyses were performed on postmenopausal women dichotomized above/below age group 50th percentile for body fat percent. TEE was reduced by 10% for the thermogenesis of digestion wherein AEE was calculated by subtracting REE from adjusted TEE. PAL was calculated as the ratio of TEE/REE. AEE was divided by the mean net energy expenditure of nongraded walking to calculate the ARTE_walk index. Between-group differences were not detected for AEE or PAL. However, the ARTE_walk index revealed that participants with less adiposity were more physically active (258 ± 149 vs. 198 ± 115 min·day⁻¹; P = 0.046; g = 0.46). AEE and PAL did not correlate with cardiorespiratory fitness or biomarkers of cardiometabolic health. Cardiorespiratory fitness (r = 0.32), arterial elasticity (r = 0.24), total cholesterol/HDL-c ratio (r = −0.22), and body fat% (r = −0.24) were correlated with ARTE_walk. The ARTE_walk index may offer utility in detecting possible differences in physical activity volume among postmenopausal women and appears better associated with cardiometabolic biomarkers compared with AEE or PAL.

INTRODUCTION

Advancing age is characterized by multiple physiological changes that can interfere with activities of daily living thereby increasing vulnerability to cardiometabolic disease. Habitual physical activity, on the other hand, is a known lifestyle-behavioral strategy for primary disease prevention and preservation of overall well being (1). Given that physical activity often declines with age (2), it is of interest to clarify the relationship between physical activity volume (i.e., daily quantity) and related health biomarkers. Recent technological advances have contributed to widespread use of accelerometry for physical activity monitoring; however, emerging evidence indicates that accelerometers possess limited generalizability (3, 4) and lack sensitivity to detect changes in the energetic cost of walking (5). More specifically, accelerometers perform poorly at slower walking speeds, and given the between-person differences in walking economy, accelerometer-derived data are ill-suited to make direct comparisons. To this end, connecting physical activity volume with clinically appropriate endpoints including cardiorespiratory fitness (V̇o_2max) and biomarkers of cardiometabolic health necessitate suitable measurement sensitivity and specificity.

Doubly labeled water represents the gold-standard technique for measurement of free-living total energy expenditure (TEE) (6) wherein activity-related energy expenditure (AEE) can be determined by accounting for the thermogenesis of digestion and subtracting resting energy expenditure (REE) from (adjusted) TEE. Though AEE reflects the amount of energy used to perform physical activity under free-living conditions (7), between-person differences in body size (8), and walking economy (i.e., net V̇o₂ calculated from steady-state work V̇o₂ minus resting V̇o₂) (9) can introduce systematic confounding. For instance, individuals with greater adiposity and/or those with poorer walking economy can exhibit a disproportionately high AEE with respect to actual physical activity volume, thus making interpretation about physical activity volume and cardiometabolic health problematic in individuals with existing comorbidities.

The TEE-to-REE (TEE/REE) ratio yields a physical activity level (PAL) index that represents relative physical activity energy expenditure (10), yet it similarly does not account for variance in walking economy. Given that the standard deviation (SD) for walking economy commonly varies by 10% or more for a standardized task (8, 11), PAL may underestimate or overestimate physical activity volume in certain individuals. For instance, approximated distance traveled for an individual with poor walking economy (i.e., −2 SD) might be ≈20% above the mean (for physical activity), whereas an individual with a good walking economy (i.e., +2 SD) might be ≈20% below the mean. Stated differently, if the actual distance traveled were 1.0 mile, we would presume the first person to have covered 1.2 miles while the second person might have covered just 0.8 miles. Despite performing the same task, these are certainly much different estimates of physical activity volume. The consequence being an inability to accurately connect physical activity volume with specific endpoints, namely, parameters of cardiometabolic health. Resolving this issue would readily enhance our ability to accurately define evidenced-based public health initiatives aimed at promoting sufficient daily physical activity.

With respect to younger adults, waning skeletal muscle strength and cardiorespiratory fitness can unfavorably affect walking economy with advancing age such that greater variability in walking economy would be expected in older adults (12, 13). To this end, the activity-related time equivalent (ARTE) was developed to account for between-person differences in body mass and walking economy by determining the ratio of AEE (kcal/day) to locomotion economy (kcal/min) during a set of activities that mimicked independent living (11, 14–16). Such reference activities included an economy composite by incorporating nongraded walking, graded walking, simulated grocery carry, stair climbing, and cycling. However, given that walking is the primary mode of physical activity for most older adults, an attractive alternative would be to simplify the determination of ARTE measure by incorporating nongraded walking alone (ARTE_walk) in lieu of the five-activity composite.

Although AEE is related to physical activity volume, these terms are not synonymous as it is more energetically costly for individuals with greater adiposity and/or lower-extremity dysfunctions to ambulate. Such instances can exaggerate AEE-derived values thereby making it difficult to reconcile emergent discrepancies among individuals of varying body fat percentage and/or walking economy. To our knowledge, comparisons involving AEE, PAL, and the ARTE_walk index in a cohort of older adults have not been performed. Herein we compared AEE, PAL, and the ARTE_walk index among postmenopausal women dichotomized above/below age group 50th percentile for total body fat percent (17). Groups were dichotomized as proof-of-principle to evaluate how group differences in body fat percentage and walking economy may differentially affect measures of AEE, PAL, and the ARTE_walk index. We also sought to explore possible associations of AEE, PAL, and the ARTE_walk index with cardiorespiratory fitness and other biomarkers of cardiometabolic health. Based on the premise that physical activity volume, not just energy expenditure, is more tightly linked to cardiometabolic health, we hypothesized the ARTE_walk would exhibit stronger associations compared with AEE and PAL.

METHODS

Design and Participants

Secondary analyses were performed on preintervention, baseline data (n = 81) from a study designed to evaluate exercise frequency in postmenopausal women (18). All participants were self-reported nonsmokers and exercised less than one time per week. Consistent with guidelines set forth by the Declaration of Helsinki study, all procedures were approved by the local institutional review board, and written informed consent was obtained from every participant.

Blood Draw and Lipid Profile

An antecubital blood draw was performed in the morning hours following an overnight fast. Consistent with the manufacturer’s instruction, serum total cholesterol (TC), high-density lipoprotein (HDL-c), and triglycerides (TG) were measured using an automated analyzer (Stanbio Laboratory, Boerne, TX). Low-density lipoprotein (LDL-c) was calculated using the Friedewald formula (FF): LDL-c (mg/dL) = TC (mg/dL) − HDL-c (mg/dL) − TG (mg/dL)/5 (19).

Resting Energy Expenditure

REE was measured using indirect calorimetry (Sensor Medics, Yorba Linda, CA) in a temperature-controlled (20°C–22°C) room. After an overnight fast, measurements were performed in the Physical Activity Core Laboratory during the morning hours. Instrumentation was calibrated before each assessment using standard gases. After an initial 10-min equilibration period, a ventilated hood was used to collect expired air for 20–30 min where oxygen uptake (V̇o₂) and carbon dioxide production (V̇co₂) were continuously measured. Energy expenditure was calculated using the equation from Weir (20).

Arterial Elasticity

Arterial elasticity was estimated from the radial pulse via contour analyses (CR-2000, Hypertension Diagnostics, Eagan, MN). Briefly, participants were in the seated position for at least 5 min wherein a solid-state pressure transducer was secured over the radial pulse of the dominant arm. A stabilizer was used to support the wrist to minimize inadvertent artifact/movement. An automated oscillatory blood pressure cuff was placed on the contralateral arm. Once a stable measurement was achieved, a 30-s analog tracing of the radial waveform was digitized at 200 samples per second. As previously described by Cohn et al. (21), beat determination is constructed using a beat-marking algorithm that determines beginning systole, peak systole, onset of diastole, and end diastole during the 30-s period. According to the manufacturer’s details, arterial elasticity estimates are based on a modified Windkessel model that permits evaluation of the conduit and microcirculatory arteries. All assessments were performed in triplicate and averaged for analyses.

Total and Activity-Related Energy Expenditure

Doubly labeled water was used to measure free-living TEE (11). Participants provided a baseline urine sample, after which, a loading dose of doubly labeled water (10% H₂¹⁸O and 8% ²H₂O) was administered at a ratio equating to ∼1 g/kg body mass. To permit suitable isotopic dilution, two additional urine samples were collected at +3 and +4 h post doubly labeled water dosing. Two urine samples were collected in the morning hours 14 days later. Urine samples were stored at −20°C and evaluated in duplicate via isotope ratio-mass spectrometry. The coefficient of variation for repeated measures in our laboratory is 4.3% (18). Carbon dioxide rates (rCO₂) were determined using a fixed constant for the dilution space ratio (1.0427) provided by Speakman (22) and converted to energy expenditure using the equation by Weir (20): TEE (kcal/d) = 3.9 (rCO₂/FQ) + 1.1 rCO₂; where rCO₂ is the rate of CO₂ production (L/d) and food quotient. Free-living AEE was determined by initially reducing TEE by 10% to account for the thermogenic effects of digestion. AEE was then calculated as the difference in REE from the adjusted TEE [AEE = (TEE × 0.9) – REE].

Physical Activity Measures

Consistent with standard practices, the PAL index was defined as the quotient of TEE/REE (23). The ARTE_walk index was calculated to examine free-living physical activity in minutes per day using the following equation: ARTE_walk index (min/day) = AEE (kcal/day)/AEC (kcal/min); where AEC is the average net energetic cost measured via indirect calorimetry during final 2 min of fixed speed (0.89 m/s) treadmill task at 0% gradient lasting between 4 and 5 min. This approach permits evaluation of between-group differences by overcoming the confounding variance in body mass and/or walking economy (11, 14–16).

Maximal Exercise Testing

A graded exercise test using a modified Balke treadmill protocol, coupled with indirect calorimetry (Physio-Dyne Instrument Corporation, Quogue, NY) was used to measure cardiorespiratory fitness. Monitoring consisted of a 12-lead electrocardiogram, whereas blood pressure measurements were performed from the brachial artery at 2 min intervals. Testing commenced with a treadmill speed of 0.89 m·s⁻¹ for 2 min. Grade was increased 3.5% every 2 min until the 12th minute, wherein grade was decreased to 12% and speed was increased to 1.34 m·s⁻¹. Further workload increases were made by increasing grade by 2.5% each minute until exhaustion. Heart rate (HR) and oxygen uptake were recorded during the last 20 s of each stage. Participants were encouraged to exercise to volitional exhaustion. Cardiorespiratory fitness (i.e., V̇o_2max) was represented by the highest 20 s average. If a plateau in V̇o₂ was not present, additional criteria for achieving V̇o_2max were 1) HR within 10 beats of age-predicted (220-age) maximum and 2) ≥RER 1.10.

Body Composition

Total body fat percent was measured by dual-energy X-ray absorptiometry (GE Lunar, Madison, WI). Participants wore light clothing and remained supine in compliance with standard testing procedures. Scans were analyzed with Adult software v1.33.

Statistical Analyses

Data were assessed for normal distributions using the Shapiro–Wilk test. Independent samples t tests were used to compare between-group differences in participants above/below age group (e.g., 42.4% for 60–69, 41.2% for 70–79 yr of age) 50th percentile for body fat percent (17). Data are presented as means and standard deviations. Effect size calculations were performed with Hedges’ g to qualify substantive differences: small (0.2), medium (0.5), and large (0.8). Pearson-product moment correlations were used to examine possible associations among AEE, PAL, and ARTE_walk with cardiometabolic health measures including cardiorespiratory fitness_, arterial elasticity, body fat percentage, and blood lipid biomarkers. After significant bivariate correlations, multiple linear regression (enter method) was used to examine the independent effects of V̇o_2max, body fat percentage, (submaximal) walking HR, large arterial elasticity, TG-to-HDL-c (TG/HDLc) ratio, and TC-to-HDL-c (TC/HDL-c) ratio on ARTE_walk index. Collinearity of diagnostics for variables was within acceptable limits with variable inflation factors for each analysis less than 1.53. Analyses were performed using the Statistical Package for the Social Science (SPSS v27.0; IBM, NY). Statistical significance for all tests was set a priori and defined as a two-sided P value ≤ 0.05.

RESULTS

Descriptive variables are shown in Table 1. Consistent with group allocation, significant between-group differences were present for body mass and fat percentage. As anticipated, between-group differences were detected for V̇o_2max. Participants above and below age group 50th percentile for body fat percent were ∼51% and 33% higher, respectively, than the requisite minimum V̇o_2max for functional independence in older women (24). HR and walking EE were significantly lower among lean participants, indicating less physiological strain for the standardized treadmill task. Resting HR and large arterial elasticity were significantly different between groups, both of which, were more favorable in participants with lower body fat percentages. However, apart from significant differences in HDL-c, all other lipid parameters including TC, TR, and LDL-c were not different between groups.

Table 1.

Descriptive characteristics

Variables	BF% < 50th Percentile	BF% > 50th Percentile	P Value	ES
	n = 34	n = 47
Age, yr	65 ± 4	64 ± 4	0.198	0.29
Height, cm	166 ± 4	164 ± 6	0.066	0.42
Body mass, kg	65.7 ± 8.4	78.6 ± 8.4	<0.001	1.51
Body fat, %	37.1 ± 4.8	47.0 ± 2.7	<0.001	2.55
V̇o2max, mL·kg−1·min−1	25.5 ± 4.6	21.0 ± 3.8	<0.001	1.07
Walking HR, beats/min^	98 ± 14	108 ± 13	0.002	0.70
Walking EE, kcal·min−1^	2.4 ± 0.5	2.9 ± 0.6	<0.001	0.84
Resting HR, beats/min	61 ± 6	65 ± 9	0.022	0.55
Systolic blood pressure, mmHg†	127 ± 14	128 ± 16	0.627	0.11
Diastolic blood pressure, mmHg†	70 ± 9	70 ± 10	0.963	0.01
Large artery elasticity, mL/mmHg·10†	13.8 ± 4.2	11.8 ± 3.2	0.027	0.53
Small artery elasticity, mL/mmHg·100†	3.3 ± 1.4	4.1 ± 2.0	0.063	0.44
TC, mg/dL	225 ± 34	212 ± 46	0.147	0.33
TG, mg/dL	106 ± 43	120 ± 52	0.201	0.28
HDL-c, mg/dL	72 ± 19	61 ± 18	0.009	0.59
LDL-c, mg/dL	132 ± 30	127 ± 37	0.508	0.15
TG/HDL-c	1.7 ± 1.1	2.3 ± 1.5	0.064	0.42
TC/HDL-c	3.3 ± 1.0	3.7 ± 1.1	0.102	0.37
LDL-c/HDL-c	2.0 ± 0.8	2.3 ± 0.9	0.145	0.33

Values are shown as means and standard deviation. BF, body fat percentage; EE, energy expenditure; EF, effect size as determined by Hedges’ g; HDL, high-density lipoprotein; HR, heart rate; LDL, low-density lipoprotein; V̇o_2max, maximal oxygen uptake assessed during a graded exercise test; TC, total cholesterol; TG, triglycerides. ^Performed during nongraded walking on a treadmill at 0.89 m/s; †Sample size for 30 and 45 in participants < and > 50^th percentile for body fat percentage, respectively. P values correspond with between-group comparison using an independent samples t test.

Table 2 shows between-group comparisons on measures of free-living PA. Significant between-group differences were not observed in AEE or PAL. Alternatively, the ARTE_walk index revealed a significant between-group difference in PA volume. Leaner participants performed 26% more physical activity when quantified using ARTE_walk, whereas between-group variations for AEE and PAL were just 8% and 5%, respectively.

Table 2.

Measures of free-living physical activity

Variables	BF% < 50th Percentile n = 34	BF% > 50th Percentile n = 47	P Value	ES
AEE, kcal/day	600 ± 320	553 ± 294	0.495	0.15
PAL, TEE/REE	1.68 ± 0.31	1.60 ± 0.27	0.206	0.28
ARTEwalk, min/day	258 ± 149	198 ± 115	0.046†	0.46

AEE, activity-related energy expenditure; ARTE_walk, activity-related time equivalent; ES, effect size as determined by Hedges’ g; PAL, physical activity index calculated as the ratio of total energy expenditure (TEE) to resting energy expenditure (REE). Data are presented as means ± standard deviation. †Statistical significance between groups.

Table 3 depicts a correlation matrix among selected variables of interest. Significant negative associations were observed between the ARTE_walk index and the following: body fat percent (r = −0.238, P = 0.032), TG/HDL-c ratio (r = −0.218, P = 0.050), and TC/HDL-c (r = −0.223, P = 0.045). Significant positive associations were observed with the ARTE_walk index and V̇o_2max (r = 0.321, P = 0.004) and large arterial elasticity (r = 0.242, P = 0.037). Notably, AEE and PAL were not associated with any fitness, body composition, or blood lipid variable. Multiple linear regression showed a higher V̇o_2max (partial r = 0.224, P = 0.050) and lower HR responses (partial r = −0.245, P = 0.033) during nongraded walking tended to be the most meaningful correlates of the ARTE_walk index (Table 4).

Table 3.

Correlation matrix among selected variables (n = 81)

Variables	1	2	3	4	5	6	7	8	9	10	11
AEE
PAL	0.956‡
ARTEwalk	0.895‡	0.888‡
Body fat	−0.042	−0.077	−0.238*
V̇o2max	0.204	0.188	0.321‡	−0.526‡
Walking HR^	−0.154	−0.152	−0.299‡	0.336*	−0.204
Large artery elasticity†	0.146	0.148	0.242*	−0.204	0.240*	−0.275*
Small artery elasticity†	0.148	0.007	0.049	0.188	−0.078	0.046	0.136
TR/HDL-c ratio	−0.166	−0.216	−0.218*	0.214	−0.123	0.059	−0.145	0.214
TC/HDL-c ratio	−0.146	−0.206	−0.223*	0.211	−0.222*	0.028	−0.130	0.099	0.848‡
LDL-c/HDL-c ratio	−0.130	−0.188	−0.210	0.196	−0.241*	0.016	−0.117	0.054	0.742‡	0.985‡

^‡Significance at P < 0.01; *significance at P ≤ 0.05.

AEE, activity-related energy expenditure; ARTE_walk, activity-related time equivalent; HDL-c, high-density lipoprotein; HR, heart rate; LDL-c, low-density lipoprotein; PAL, physical activity index calculated as the ratio of total energy expenditure/resting energy expenditure; V̇o_2max, maximal oxygen uptake assessed via graded exercise test; TC, total cholesterol; TR, triglycerides. ^Response during a nongraded, fixed speed (0.89 m/s) treadmill task; †n = 75.

Table 4.

Model estimation for physical activity behavior indexed by the activity-related time equivalent (n = 81)

	Model R	R 2	Partial r	P Value
Model 1: ARTEwalk	0.40	0.16
V̇o2max			0.234*	0.039
Body fat%			−0.021	0.852
Walking HR			−0.237*	0.037
Model 2: ARTEwalk‡	0.41	0.17
V̇o2max			0.250*	0.033
Walking HR			−0.191	0.106
Large artery elasticity			0.134	0.258
Model 3: ARTEwalk	0.44	0.19
V̇o2max			0.224*	0.050
Body fat%			0.011	0.925
Walking HR			−0.245*	0.033
TR/HDL-c ratio			−0.073	0.529
TC/HDL-c ratio			−0.031	0.789

^‡n = 75 in model. HDL-c, high-density lipoprotein; HR, heart rate responses during a nongraded treadmill task (0.89 m·s⁻¹); TC, total cholesterol; TR, triglycerides; V̇o_2max, maximal oxygen uptake measured via indirect calorimetry during treadmill graded exercise test. *P value less than or equal to 0.05.

DISCUSSION

Habitual physical activity positively associates with cardiometabolic health; however, the strength of this principle is reliant on the ability to correctly distinguish such activity/behavior with measurable, clinically relevant endpoints. Here, we present evidence that variation in walking economy can obscure the sensitivity for detecting possible differences in physical activity volume among postmenopausal women using the more traditional measures of AEE and PAL. These findings are not meant to suggest that AEE or PAL are ineffective rather our findings underscore the value of normalizing AEE by accounting for individual-specific walking economy. This approach may be especially important when examining intervention effects including weight loss and/or exercise training, both of which are known to affect walking economy (8).

Indeed, AEE is not synonymous with physical activity volume as it is more energetically costly for individuals with greater adiposity and/or lower-extremity dysfunctions to ambulate. This, in turn, can exaggerate AEE-derived values, which make it difficult to reconcile differences emerging among individuals who vary by body fat percentage and/or walking economy. Since larger individuals tend to have greater REEs, it is widely accepted that PAL (TEE/REE) accounts for confounding effects of body size. However, PAL does not adjust for differences in walking economy. A natural dispute may be to question what dependent variable is of primary importance. Does physical activity volume or energetic flux matter more in relation to cardiometabolic health? This is not something we answer here, and understandably, is a debate that will persist for the foreseeable future. However, based on present results, we contend the ARTE_walk index may offer utility in detecting possible differences in physical activity volume in postmenopausal women and appears better associated with cardiometabolic biomarkers compared with AEE or PAL.

The ARTE index was developed over 20 years ago in response to noted differences in locomotion economy between African American and European American women (11). In that study, Weinsier et al. (11) made comparisons on measures of V̇o_2max and AEE before and after diet-induced weight loss. Results showed AA women had a better overall economy as evidenced by a 6% lower energetic cost of locomotion. Such differences were the rationale behind normalizing AEE to a locomotion economy composite developed from five physical tasks meant to simulate independent living. Further observations showed a modest 3% difference in AEE between races after weight loss, however, adjustment for locomotion economy (i.e., ARTE index) expanded to 11%. Though these differences were not statistically significant, such findings reveal how individual-specific locomotion economy can affect inferences about free-living physical activity behavior. More recent observations have shown an inverse relationship between V̇o_2max and walking economy (8), such that changes in V̇o_2max and walking economy were significantly related (independent of fat-free mass) with weight loss. The implications being that walking economy may not be a fixed construct (5) and should be measured when testing intervention effects and/or performing longitudinal analyses of free-living physical activity. Though speculative, it is conceivable, other factors linked to walking economy, including variation in adiposity, skeletal muscle strength, joint stiffness, and stretch-shortening cycle potentiation (13) could obscure inferences about AEE.

Whereas a physically active lifestyle generally corresponds with favorable health and wellness, linking such behavior with clinically relevant endpoints necessitates measurement sensitivity and specificity. In the present work, we have shown between-group differences using the ARTE_walk index but not AEE or PAL. Though we cannot definitively confirm which physical activity measure was more accurate, it is notable the ARTE_walk index (not AEE or PAL) significantly correlated with several biomarkers of cardiometabolic health including V̇o_2max large arterial elasticity, TR/HDL-c ratio, TC/HDL-c ratio, and body fat percent. Given these findings are correlative in nature, we can only speculate about the meaningfulness of these relationships. Still the implications being the ARTE_walk index may be a sensitive measure of physical activity volume—a known construct readily involved in attenuating risk of cardiometabolic dysregulation (25).

Largely because of reliance on integrating a measure of walking economy, the ARTE_walk index may be well suited to examine physical activity volume in response to exercise training, weight loss, and/or natural aging. Even subtle factors affecting the energetic cost would be reflected in the ARTE_walk index. In addition to slow preferred walking speed, exhibiting a higher energetic cost of walking is a known predictor of multiple health-related complications including disability, cognitive decline, and mortality (26, 27). This, in turn, highlights the unique function of the ARTE_walk index to evaluate free-living PA among older adults and/or clinical populations.

In the present work, the PAL (TEE/REE) index did not make a between-group distinction in physical activity. To correct for variations in physical activity, REE would have to strongly relate to the energetic cost of locomotion (or simply walking). This may be an important feature among individuals with advancing age wherein a large variability exists in the energetic cost of walking. Here, one standard deviation in the energetic cost of walking for the total sample was 14%, meaning some participants were 28% more energetically costly (i.e., 2 SDs above the mean) despite performing the same treadmill task. Indeed, REE was not related to the energetic cost of walking during nongraded treadmill walking (r = 0.103; P = 0.364). Thus, it is unsurprising in this instance that PAL appears insensitive to variations in V̇o_2max.

There are several limitations of the present work that should be considered. Inherent with the cross-sectional design we cannot establish directionality in the reported relationships. Future work should employ assessments of the ARTE_walk index over time as well as calculating AEC using a “preferred” walking speed. The latter point may offer broader utility to clinical populations affected by joint pain and/or lower-extremity joint dysfunction, which are known to increase the energetic cost of walking. The results are restricted to postmenopausal women without a history of cardiometabolic disease, such that extrapolation to other demographics should be performed with caution. Thus, research is needed to evaluate the application of the ARTE_walk index in other demographics, especially those affected by chronic disease or restricted mobility. As indicated previously, the present results are not necessarily suggestive that ARTE_walk is superior to AEE or PAL, rather that it may be more sensitive to physical activity volume among heterogeneouss groups with large variation in walking economy. Still, it is noteworthy that the ARTE_walk index was correlated with several biomarkers of cardiometabolic health that were not apparent for AEE or PAL. Strengths of the current work are the use of doubly labeled water to measure free-living TEE, indirect calorimetry to measure walking economy, and V̇o_2max as well as the inclusion of lipid biomarkers and arterial elasticity.

Perspectives and Significance

We show evidence that variation in walking economy can obscure the sensitivity for detecting possible differences in physical activity volume among postmenopausal women. Though we cannot determine if AEE, PAL, or ARTE_walk index is more accurate at measuring physical activity, it is notable the ARTE_walk index (but not AEE or PAL) significantly correlated with multiple biomarkers of cardiometabolic health including V̇o_2max, large arterial elasticity, TR/HDL-c ratio, TC/HDL-c ratio, and body fat percent. The implications of these findings support the premise that the ARTE_walk index may be uniquely suited to evaluate free-living physical activity among older adults and/or clinical populations where concerns over variation in walking economy may be significant.

GRANTS

The present work was supported by the following National Institutes of Health (NIH) Grants: R01AG027084 (to G. R. Hunter), R01AG027084-S (to G. R. Hunter), and P30DK056336. S. J. Carter also receives support from the Indiana Clinical and Translational Sciences Institute funded by Grant UL1TR002529 from the NIH, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award. This study is also supported by ClinicalTrials.gov Identifier: NCT01031394 (to G. R. Hunter).

DISCLOSURES

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

AUTHOR CONTRIBUTIONS

S.J.C. and G.R.H. conceived and designed research; G.R.H. performed experiments; S.J.C., M.N.B., and G.R.H. analyzed data; S.J.C., M.N.B., H.S., C.M., and G.R.H. interpreted results of experiments; S.J.C., M.N.B., H.S., C.M., and G.R.H. prepared figures; S.J.C., M.N.B., H.S., C.M., and G.R.H. drafted manuscript; S.J.C., M.N.B., H.S., C.M., and G.R.H. edited and revised manuscript; S.J.C., M.N.B., H.S., C.M., and G.R.H. approved final version of manuscript.