Accelerometer-Determined Intensity and Duration of Habitual Physical Activity and Walking Performance in Well-Functioning Middle-Aged and Older Women: A Cross-Sectional Study

Abstract

Background

The association of physical activity (PA) intensities and duration spent in those activities with different walking tasks remains unclear.

Objectives

To examine the association between the duration of PA intensities and three walking speeds (usual walking speed, maximal walking speed and zig-zag walking speed).

Design

Multiple linear regression analysis was used to estimate the association of age, BMI, maximum knee extension strength, light PA, moderate PA and vigorous PA with walking speeds. Setting: University lab. Participants: Eighty-six older women (67 ± 7 years).

Measurements

PA was measured for 30 consecutive days using the Lifecorder-EX accelerometer. Exercise intensity was categorized as light (levels 1-3), moderate (levels 4-6) and vigorous (levels 7–9) based on the manufacturer algorithms. Usual straight walking speed (20 m), maximal straight walking speed (20 m) and zig-zag walking speed tests (10 m) were performed by each participant.

Results

For the usual straight walking speed model (R² = 0.296, SEE = 0.15 m/s), the significant predictors were BMI, knee extension strength, light PA and vigorous PA. For the maximal straight walking speed model (R² = 0.326, SEE = 0.20 m/s), only age was a significant predictor. For the zig-zag walking speed model (R²=0.417, SEE = 0.14 m/s), age and maximum knee strength were significant predictors in the model.

Conclusions

Overall, the results of this study suggest that vigorous PA and maximal knee extension strength are two important factors that are associated with different walking speeds in older women.

Abbreviations

PA

Physical Activity

MVC

Maximal Voluntary Isometric Contraction

BMI

body mass index

S.β.

Standardized beta coefficient

Introduction

Walking speed is a powerful predictor of disability (1, 2), hospitalization (2, 3) and mortality (4, 5) in older adults, regardless of chronic health conditions. Due to the decline in walking speed being associated with various negative health outcomes, researchers have investigated the association of strength training and daily ambulatory activity on attenuating the declines in usual and maximal walking speed. However, the association of strength training on walking speeds is inconsistent (6, 7, 8, 9). Interestingly, daily ambulatory activity may play a bigger role. Cross-sectional studies report that accelerometer-determined habitual physical activity (PA) is associated with usual and maximum walking speed in older adults (10, 11, 12, 13, 14). Nevertheless, little is known about how the intensity and duration of daily ambulatory activity may be associated with walking performance in older adults. Additionally, except for one study (10), all the other studies measured only one week (7 days) of daily PA using an accelerometer or a pedometer (11, 12, 13, 14, 15). In order to achieve >90% reliability for estimating the yearly based PA using step counts in older women, at least 30 consecutive observation days are required (16).

The purpose of this study was to identify if the intensity (light, moderate and vigorous) and duration of accelerometer-based habitual PA measured for 30 consecutive days were associated with three different types (usual and maximum straight and zigzag) of walking speeds in middle-aged and older women. The accelerometer recognizes changes in acceleration which can be classified from light to vigorous PA. It is possible that many opportunities for vigorous daily activities may stimulate better walking performance in older adults. Thus, we hypothesized that walking speed would be associated with the duration of moderate to vigorous PA during daily life.

Materials and Methods

Participants

The participant data (n=86, age range 51–83 years) came from a previous study investigating a different topic (17) and a new data set was added where the participants completed the same procedures as the other data set. All participants from the two data sets were recruited from the Toyo Gakuen university campus and surrounding area through printed advertisement and by word of mouth. Exclusion criteria included being under the age of 50 years, serious medical illness, a history of psychosis or depression, alcohol and drug abuse within the past 12 months. Prior to obtaining informed written consent to participate in each respective study, a written description of the purpose of the study and its safety was distributed to potential participants. All participants were free of chronic health conditions as assessed by self-report. Approximately 70% of participants reported regular sports activity (at least once a week) which included structured walking activity. In addition, our participants were well-functioning. The criteria used to define well-functioning was a usual walking speed >1.0 m/s and a handgrip strength >18 kg. This study was conducted according to the World Medical Association Declaration of Helsinki and was approved by the University's Ethics Committee for Human Experiments.

Body mass index

Body mass and standing height were measured to the nearest 0.1 kg and 0.1 cm, respectively, by using a height scale (YL-65, Yagami Inc., Nagoya, Japan) and an electronic weight scale (WB-150, Tanita, Tokyo, Japan). Body mass index (BMI) was defined as body mass divided by height squared (kg/m²).

Accelerometer-determined daytime PA

Habitual PA was measured using a uniaxial accelerometer (Lifecorder-EX, Suzuken Co. Ltd., Nagoya, Japan) for 30 consecutive days. Prior to that, participants attached the accelerometer on a customized belt over their right hip and were instructed to record their daily PA for 30 consecutive days. The accelerometers were worn at least 12 hours each day. If the participants did not wear the accelerometer for two consecutive days during the testing period, they were requested to additionally wear the accelerometer for the numbers of days missed and to record that information.

The accelerometer measures acceleration in the vertical direction (z-axis) every four seconds with a piezoelectric ceramic accelerometer. If three or more acceleration pulses were detected (≥0.06 G) during four consecutive seconds, then the activity was classified into one of nine activity levels (1, 2, 3, 4, 5, 6, 7, 8, 9) based on the manufacturer's algorithm which is proprietary (18). When activity was less than 0.06 G, then the level was classified as 0 denoting immobility. Only levels 1–9 were printed on the reports given to participants that were used for data collection. After the measurements were taken, the recorded activity level data were downloaded to a personal computer. In this study, exercise intensity was categorized into 1 of 3 activity levels based on the accelerometer signal as described previously (18): light-PA (levels 1–3, <3 METs), moderate-PA (Levels 4–6, 3–6 METs) and vigorous-PA (levels 7–9, >6 METs). In addition, the total duration of each level of exercise intensity was calculated. Average daily step count was also measured.

Gait performance

Usual straight walking, maximum straight walking and maximum zigzag walking measured gait performance. The total length of the marked corridor was 24-m and the width of the corridor was one meter. Two-meter acceleration and deceleration zones were used for usual and maximum walking tests (20-m timed distance). Participants were first asked to walk at their usual (walk at a normal pace) speed and were then invited to perform the maximum (as fast as possible without running) walking speed test. Zig-zag walking time was measured using a 10-m walkway. Four cones were placed 2-m apart on the floor between the start and finish points (19). The cones were set to alternate from side to side with a distance of 0.5-m from a line drawn through the start and finish points. The test started with the participant standing on the start point. The participant then walked around the outside of each cone and walked through the finish point. They were asked to walk as quickly as possible. The zig-zag walking was selected to evaluate a relatively difficult task performance which is associated with site-specific thigh muscle loss (19). Participants performed two timed trials for each test. Times were measured using a digital stopwatch (Seiko, ADMD-001). The best time was converted to a maximum speed measurement (m/s) for the maximum straight and zigzag walking speed and average value of the two trials was used for the usual walking speed.

Maximal isometric strength

Maximum isometric knee extension strength was measured using a Biodex System-3 dynamometer with the hip angle at 85° and arms crossed across the chest (Shirley, New York, USA). Participants were familiarized approximately one week before testing. The center of rotation of the knee joint was visually aligned with the axis of the lever arm of the dynamometer, and the ankle of the right leg was firmly attached to the lever arm of the dynamometer with a strap. Three warm-up submaximal contractions were done and then the subjects performed maximal voluntary isometric contraction (MVC) knee extension at a knee joint angle of 80° (0° corresponded to full extension of the knee). If MVC strength for the first two trials varied by >5%, an additional MVC was performed. The coefficient of variation for this test in our laboratory was 7%.

Statistical analysis

Multiple linear regression models using the enter method in SPSS were used to estimate the association of different variables with usual straight walking speed, maximum straight walking speed and zigzag walking speed. Variables included were age, BMI, maximum knee extension strength, light PA, moderate PA and vigorous PA. Linearity and homogeneity of variance assumptions were tested by plotting standardized residuals vs. standardized predicted values. Normality of residuals was checked using Q-Q plots. Multicollinearity was checked using the variance inflation factor (VIF) statistic and multicollinearity was considered violated if VIF values were greater than 10. Statistical analyses were performed using IBM SPSS statistics 24. Statistical significance was set at an alpha level of 0.05.

Results

Participant characteristics are found in Table 1. Linearity and homogeneity of variance assumptions were confirmed from residual plots. Normality of residuals was confirmed with Q-Q plots and multicollinearity was not present with the VIF statistic < 10 for the different predictors. The physical activity durations for different intensities were 60 (SD 19) min/day for light, 21 (SD 14) min/day for moderate and 2 (SD 2) min/day for vigorous activity. The model for usual straight walking speed was significant [F (6, 79) = 5.540, p<0.001] with an R² of 0.296 and an adjusted R² value of 0.243. Significant predictors for usual straight walking speed included BMI (p=0.037), knee extension strength (p=0.002), light PA (p=0.016) and vigorous PA (p=0.003) (Table 2). The most associated variable with usual walking speed was vigorous PA (standardized beta coefficient (S.β.) = 0.357).

Table 3.

Maximal straight walking speed model. BMI = body mass index, KE = knee extension strength, PA = physical activity, VIF = variance inflation factor

Variables	Beta Coefficients	Standardized Beta Coefficients	t-value	p-value	VIF
(Constant)	3.218		8.713	<0.001
Age (years)	−0.013	−0.368	−3.557	0.001	1.251
BMI (kg/m2)	−0.017	−0.179	−1.836	0.070	1.118
KE Max (Nm)	0.001	0.108	1.070	0.288	1.184
Light-PA (min/day)	−0.001	−0.061	−0.615	0.540	1.164
Moderate-PA (min/day)	0.001	0.031	0.273	0.785	1.498
Vigorous-PA (min/day)	0.023	0.199	1.768	0.081	1.483

Table 1.

Participant characteristics. U-walk = usual straight walking speed, M-walk = maximal straight walking speed, Z-walk = zig-zag walking speed, BMI = body mass index, KE = knee extension strength, PA = physical activity

Characteristics	Mean	Standard Deviation
U-walk speed (m/s)	1.56	0.18
M-walk speed (m/s)	2.10	0.23
Z-walk speed (m/s)	1.51	0.17
Age (years)	67	7
BMI (kg/m2)	22.4	2.5
KE Max (Nm)	101.3	24.5
Light-PA (min/day)	60.1	18.9
Moderate-PA (min/day)	21.2	14.0
Vigorous-PA (min/day)	1.9	2.0
Step Counts	7930	2769

Table 2.

Usual straight walking speed model. BMI = body mass index, KE = knee extension strength, PA = physical activity, VIF = variance inflation factor

Variables	Beta Coefficients	Standardized Beta Coefficients	t-value	p-value	VIF
(Constant)	1.796		6.223	<0.001
Age (years)	0.000	−0.014	−0.134	0.894	1.251
BMI (kg/m2)	−0.015	−0.212	−2.120	0.037	1.118
KE Max (Nm)	0.002	0.329	3.202	0.002	1.184
Light-PA (min/day)	−0.002	−0.250	−2.451	0.016	1.164
Moderate-PA (min/day)	−0.001	−0.112	−0.965	0.337	1.498
Vigorous-PA (min/day)	0.032	0.357	3.107	0.003	1.483

For maximal straight walking speed, the model had a R² value of 0.326 and an adjusted R² value of 0.274 [F (6,79) = 6.356, p<0.001]. The only significant predictor for maximal straight walking speed was age (p=0.001). The variable associated the most with maximal straight walking speed was age (S.β.= 0.368) followed by vigorous PA (S.β.= 0.199) and then BMI (S.β.= − 0.179).

The model for zig-zag walking had a R² of 0.417 and adjusted R² of 0.372 [F (6, 79) = 9.4, p<0.001]. Significant predictors of zig-zag walking were age and maximum knee extension strength (Table 4). Age had the greatest association with zig-zag walking speed (S.β.= −0.395).

Table 4.

Zig-zag walking speed model. BMI = body mass index, KE = knee extension strength, PA = physical activity, VIF = variance inflation factor

Variables	Beta Coefficients	Standardized Beta Coefficients	t-value	p-value	VIF
(Constant)	2.075		8.167	<0.001
Age (years)	−0.010	−0.395	−4.115	<0.001	1.251
BMI (kg/m2)	−0.007	−0.100	−1.105	0.273	1.118
KE Max (Nm)	0.002	0.219	2.340	0.022	1.184
Light-PA (min/day)	0.002	0.167	1.796	0.076	1.164
Moderate-PA (min/day)	0.000	0.021	0.201	0.841	1.498
Vigorous-PA (min/day)	0.008	0.092	0.877	0.383	1.483

Discussion

Several key findings were noted from this study. First, the amount of time spent in vigorous PA had the greatest association with usual straight walking speed while aging seemed to be playing a larger role with differences seen in maximal straight walking speeds. Second, aging and maximum knee extension strength are associated more with zig-zag walking performance then to the amount of time spent at different intensity levels of PA.

We originally hypothesized that moderate and vigorous PA levels would be associated with different walking speeds. Interestingly, only light intensity and vigorous intensity PA duration were significant predictors of usual straight walking speed. We found an inverse association between light intensity duration and usual straight walking speed such that the more time that was spent in light intensity activities, the slower the usual walking speed. The results also indicated that those who engaged in more vigorous activity throughout the day had faster usual walking speeds. This idea is supported by the study of Adachi et al. (20) which found that moderate to vigorous daily activity was inversely related to slow walking speeds in women over 75 years old. Future research may be required to determine the association of short duration vigorous PA to prevent declines in usual walking speeds over short periods of time.

For maximal walking speeds, we found that age was the only significant predictor. It is established that as people age, walking speed declines (21). However, it is unclear which factors contribute the most to this decline. The process of aging itself results in many changes to the body that could potentially be associated with maximal walking speed. For example, the loss of triceps surae muscle mass with aging is associated with decreased maximal walking speeds (22). Further, Clark et al. (23) found that the plantar flexion rate of force development was 38% lower in slower walking older adults compared to faster walking adults. They also found that the rate of neuromuscular activation of the triceps surae was 34% lower in the slower group compared to the faster group (23). These factors were not measured in the current study but could be important reasons why age was a significant predictor of maximal straight walking speeds. Zig-zag walking speed was significantly predicted by age and knee extension strength. Lateral movement and quickness may be a crucial factor during zig-zag walking. Previously, we found that zig-zag walking speed was correlated with site-specific thigh muscle loss and knee extension strength per unit body mass (19). Changes in the neuromuscular system with aging such as a decrease in motor units and a decrease in coordination of afferent and efferent action potentials could also be possible reasons why aging was associated with zig-zag walking (24).

In this study, we used uniaxial accelerometers to estimate the duration of physical activity at three different intensities (light, moderate, and vigorous). A study investigating the validity of the duration of physical activities at each intensity using uniaxial and triaxial accelerometers in older adults found no significant differences in moderate PA (3–5.9 METs) and vigorous PA (>6.0 METs) duration between the uniaxial and triaxial devices (25). Although the uniaxial accelerometer showed a longer duration for 1.1–1.9 MET activities and a shorter duration for 2.0–2.9 MET activities compared with triaxial accelerometer, the total duration of light-PA (1.1–2.9 METs) was identical between the two accelerometers. These results suggest that the estimated time duration of each PA intensity may be similar between uniaxial and triaxial accelerometers when using these cutoff values for physical activity intensity.

Several limitations exist for this study. Only 24 to 37% of the variance in walking speeds was explained by our models. Therefore, other factors, such as fitness levels, sedentary time and changes in the neuromuscular system may contribute to the variation found in gait. In addition, this was a sample size consisting of women so these results may not be applicable to men and the walking speeds were higher in our population compared to others (8). The reason for the higher walking speeds in our study is unclear but 70% of the subjects reported some type of regular sports activity which included walking. In addition, Tokyo has one of the fastest paces of life in the world which could partially explain the faster walking speeds (26). Some studies also suggest that the accelerometer cutoff points for different exercise intensities may need to be adjusted for older adults (27, 28) which may also explain why vigorous activity durations were lower in this study. However, it was not possible to adjust accelerometer cutoff points for different intensities using the accelerometer in this study due to the proprietary nature of the cutoff points. However, a strength of this study was that activity levels were assessed over a 30-day period which strengthens the reliability of these measures.

Conclusions

Overall, vigorous PA and maximum knee extension strength are important predictors for usual walking speed and zig-zag walking speed in healthy older women. When appropriate, clinicians should encourage patients to increase the amount of time spent doing vigorous activity as high as reasonably achievable (29) and increase knee extension strength as these are associated with higher walking speeds which have been associated with lower disability and chronic disease.

Conflict of Interest: The authors have no conflicts to declare.