The effect of age on the association between daily gait speed and abdominal obesity in Japanese adults

The aim of this work was to investigate the effect of age on the association between daily gait speed (DGS) and abdominal obesity defined by visceral fat area (VFA). A cross-sectional study was performed using data from an annual community-based health check-up. A total of 699 participants aged 20–88 years were enrolled in this analysis. DGS was assessed using tri-axial accelerometers worn for ≥ 7 days with at least 10 measuring hours each day. VFA was measured using a visceral fat meter. Since DGS differed significantly with age, the participants were divided into two groups: younger adults (YA), aged 20–49 years, and older adults (OA), aged 50–88 years. The association between DGS and VFA differed significantly with age (r = 0.099 for YA and r =  − 0.080 for OA; test for difference between correlation coefficients, P = 0.023). In OA, the adjusted odds ratio of abdominal obesity (VFA ≥ 100 cm2) was 0.40 (95% confidence interval 0.18, 0.88, P = 0.022) for the highest DGS quartile (DGS ≥ 1.37 m/s) compared to that for the lowest quartile (DGS < 1.11 m/s), whereas no significant association was found in YA. These data could aid in raising awareness of the self-management of obesity via DGS monitoring, especially in OA.

DGS measurement. DGS was continuously measured on a daily basis using a tri-axial accelerometer (HW-100, Kao Corporation, Tokyo, Japan) 21 . The accelerometer provides 40 days of continuous recording at a sampling frequency of 64 Hz 25 . This device detects the step cycle during gait, ranging from 70 to 160 steps/min, via medio-lateral and vertical acceleration. It commences recording tri-axial acceleration during the gait cycle if the measured acceleration of the current cycle and the two preceding cycles are within 10% of one another. Therefore, the accelerometer records tri-axial acceleration during the steady-state periods of gait. DGS was calculated using a model that used composite acceleration during one gait cycle from the tri-axial acceleration measurements. An average DGS was obtained for valid days during a 10-day period for each participant. The accelerometer also measured wearing time and step counts 25 .
The participants were instructed to wear the HW-100 on their waists at all times while they were awake, except during swimming or bathing, and to maintain their usual activities. Additionally, the participants were instructed to start wearing the HW-100 soon after their health check-up was completed and to return it after 10 days.
Measurement of other parameters. VFA was measured using a bio-impedance type visceral fat meter 24,27 , EW-FA90 (Panasonic Corporation, Osaka, Japan), which is a certified medical device in Japan (No. 22500BZX00522000) that measures VFA in a non-invasive way. The measurements obtained by this device correlate highly with those obtained using computed tomography 27 , the gold standard for VFA measurement. The following clinical characteristics were also measured: height, body weight, BMI, serum glucose, haemoglobin A1c, systolic blood pressure, diastolic blood pressure, low-density lipoprotein cholesterol (LDL cholesterol), high-density lipoprotein cholesterol (HDL cholesterol), and serum triglyceride levels. Blood samples were collected from the peripheral veins. All laboratory measurements were outsourced to LSI Medience Co. (Tokyo, Japan) and conducted according to their standard operating procedure. Data on smoking habits and medications were procured through questionnaires prepared for the health check-up. Daily intake of energy and alcohol was calculated using the Brief Diet History Questionnaire 28 .
Abdominal obesity was defined as VFA ≥ 100 cm 2 in the cross-section of umbilical level and general obesity was defined as BMI ≥ 25 kg/m 2 , according to the definition published by the Japan Society for the Study of Obesity 29 . Hypertension was defined as blood pressure ≥ 140/90 mmHg or the use of antihypertensive drugs 30 . Diabetes was defined as fasting serum glucose ≥ 126 mg/dL, HbA1c ≥ 6.5%, or the use of antidiabetic drugs 31 Dyslipidaemia was defined as LDL cholesterol ≥ 140 mg/dL, HDL cholesterol < 40 mg/dL, triglycerides ≥ 150 mg/ dL, or the use of antihyperlipidemic drugs 32 .
Statistical analysis. Participant characteristics are reported as means ± standard deviations (SDs) or percentages. Mann-Whitney U tests were used when the number of participant groups was two, whereas for comparisons of more than two participant groups, Cochran-Armitage trend tests were used. Chi-square tests and two-sample proportion tests were used for categorical variables. For comparisons between groups, Cohen's d was calculated as a measure of effect size. The overall relationship between age and DGS was evaluated using several regression models (e.g. polynomial and spline) to calculate the cut-off age for DGS decline. Smoothing spline, which is commonly used for building explanatory models in clinical research 33 , was chosen because of goodness-of-fit-based on Akaike information criterion. The spline curve was fitted with the optimal value of the smoothing parameter, determined by a generalised cross-validation method 34 . To evaluate the association between DGS, age, VFA, and BMI, we used Spearman's correlation coefficient and partial correlation coefficient and assessed the differences between the correlation coefficients. Gait speed is known to be positively associated with the number of daily steps 35 ; therefore, a partial correlation between gait speed and age was adjusted by daily steps and sex. Multiple logistic regression analysis was performed to investigate the adjusted odds ratio (aOR) and 95% confidence interval (CI) for abdominal and general obesity status, comparing participants in the higher  37 , alcohol intake 38 , and lifestyle-related diseases such as hypertension, diabetes, and dyslipidaemia 2-4 . We performed logistic regression analysis considering these factors. Model 1 was adjusted for sex and age, model 2 was further adjusted for alcohol consumption (g/day), smoking habit (cigarettes/day), and total energy intake (kcal/day), model 3 was additionally adjusted for average steps (/day), and model 4 was adjusted in a manner similar to model 3 along with incorporation of hypertension, diabetes, and dyslipidaemia. Statistical tests were two-tailed, and results with P < 0.05 were considered significant. All analyses were performed using SPSS (version 25; SPSS Inc., Chicago, IL, USA) and R (version 3.6.2; R Core Team, Vienna, Austria).

Results
The relationship between DGS and age of the study participants (n = 699) is presented in Fig. 1a. The relationship was investigated using a smoothing spline, in which the vertex was at the age of 49.8 years. The relationship between age and DGS was significantly different before (younger adults [YA]; age, 20-49 years) and after (older adults [OA]; age, 50-88 years) the vertex (r = 0.130 for YA and r = − 0.388 for OA; test for difference between correlation coefficients, P < 0.001). This difference remained significant after adjusting for sex and average steps (r = 0.091 for YA and r = − 0.323 for OA; test for difference between correlation coefficients, P < 0.001). Therefore, we divided the participants into YA and OA groups. Table 1 shows participant characteristics of the YA (n = 255, 62.4% women) and OA (n = 444, 62.2% women) groups. Mean age was 38.8 ± 6.7 years for YA and 63.6 ± 8.0 years for OA. OA had slower DGS (P < 0.001, Cohen's d = 0.30), and higher VFA (P < 0.001, Cohen's d = 0.28) and BMI (P < 0.001, Cohen's d = 0.17) than YA. The proportion of abdominal obesity was significantly higher in OA than in YA (P = 0.048), whereas the proportion of general obesity was not significantly different. No significant differences were found in the accelerometer wear time. Smoking was less prevalent in OA (P = 0.005, Cohen's d = 0.15) and this group had a higher total energy intake than YA (P < 0.001, Cohen's d = 0.24); alcohol intake was comparable. The proportions of hypertension, diabetes, and dyslipidaemia were higher in OA than in YA (all P < 0.001).
The effect of age on the association between DGS and VFA or BMI was analysed. There was a significant difference in VFA between the two groups (r = 0.099 for YA and r = − 0.080 for OA; test for difference between correlation coefficients, P = 0.023; Fig. 1b). Similar results were obtained for BMI (r = 0.067 for YA and r = − 0.124 for OA; test for difference between correlation coefficients, P = 0.015; Fig. 1c).
To evaluate the association between DGS and abdominal and general obesity, we divided the participants into quartiles according to their DGS, specifically, DGS < 1.11, 1.11 ≤ DGS < 1.23, 1.23 ≤ DGS < 1.37, and DGS ≥ 1.37. The distribution of YA and OA in the overall DGS quartiles is shown in Table 2. In OA, the proportion decreased from quartile 1 to 4, whereas it increased in YA. The proportion of OA in quartile 1 was significantly higher than that of YA (P < 0.001), and the proportion of OA in quartile 4 was significantly lower than that of YA (P = 0.034).
The aOR for the associations between DGS and abdominal or general obesity are presented in Table 3

Discussion
This is the first study to investigate the effect of age on the association between DGS and abdominal obesity, as well as general obesity. We found that the tendency of DGS was changed before and after the vertex age (50 years). Therefore, we divided the participants into two groups and found that the associations between DGS and VFA or BMI were significantly different between YA and OA (Fig. 1b,c). One possible explanation for this difference is that the energy expenditure of daily walking differs with age. A previous meta-analysis indicated that elderly participants (mean age ≥ 59 years) expend more energy than younger participants (mean age 18-41 years) when walking at comparable speeds 6 . Ortega et al. 39 demonstrated that elderly participants (mean age 76 years) expended 34% more metabolic energy while walking than younger participants (mean age 25 years). Therefore, age differences might affect the associations between DGS and VFA or BMI.
In the present study, higher DGS was associated with a significantly lower aOR of abdominal and general obesity in OA. Several studies have demonstrated that the energy cost of walking depends on speed [11][12][13] . Ortega et al. demonstrated that this dependence was stronger in elderly subjects 39 . Therefore, the present study demonstrated that faster walking might contribute to the prevention of abdominal and general obesity in OA, but not in YA. Although we adjusted for several potential variables that have been reported to be associated with obesity, other age-related variables such as gut microbiota may mediate the link between DGS and obesity through energy metabolism 40,41 . Another possible hypothesis is that a gradual loss of muscle fibres and motor units, which begins at approximately 50 years of age 42 , affects energy expenditure thereby modifying the relationship between DGS and obesity in OA.  Table 2. The distribution of younger adults and older adults in the overall daily gait speed quartiles. Equality of proportions was analysed. Each age in the four groups was compared using Cochran-Armitage trend tests. www.nature.com/scientificreports/ Several studies have reported an association between gait speed and abdominal or general obesity; however, the results were inconsistent. Ko et al. 18 reported significant associations between general obesity and decreased gait speed in subjects aged 50-84 years. Similar results were also reported by other studies on elderly subjects [14][15][16] . However, Moreira et al. 19 found no significant association between gait speed and abdominal obesity in middleaged subjects aged 40-65 years. For DGS, Schimpl et al. 20 reported no significant association between DGS and general obesity in their study population, which comprised individuals within a broad age range, of 17-65 years. One of the reasons for these inconsistent results might be that these studies did not stratify the subjects by age. In the present study, we divided our participants into two groups (YA, 20-49 years and OA, 50-88 years), and found that the association between age and DGS was significantly different between YA and OA. We also observed that DGS clearly decreased with age in OA (Fig. 1a). These data indicated that there might be an important change between the middle and older ages (e.g. a loss of muscle fibres 42 ), which caused the decline in DGS, thereby affecting the association between DGS and abdominal or general obesity.

P-value m/s % (n) % (n)
Abdominal obesity increases with age 43 , whereas gait speed decreases 20,21 . In our study participants, OA had significantly higher proportions of abdominal obesity than YA. The DGS of OA was significantly lower than that of YA, which led to a higher proportion of OA in quartile 1 (DGS < 1.11 m/s) and a lower proportion of OA in quartile 4 (DGS ≥ 1.37 m/s) compared to YA. As the world population grows older 44 , the population with lower DGS suffering from abdominal obesity might increase in the future. Therefore, DGS is expected to be one of the useful indicators for preventing abdominal obesity in people over 50 years of age. However, further studies are warranted to validate these findings.
Several studies describing the association between gait speed and health status were well summarised by Middleton et al. 45 , and gait speed was regarded as a functional vital sign. Sun et al. 46 demonstrated that women whose walking pace was brisk or very brisk (≥ 1.34 m/s) had 2.68-fold increased odds of successful aging compared to women with an easy walking pace. There are several reports for setting cut-off points, especially for elderly people. For example, there was a lower risk of events and better survival at a gait speed > 1 m/s and even better prospects for extremely fit individuals (gait speed > 1.3 m/s) 45,47 . However, it is noteworthy that these cut-off points are assessed by gait speed in laboratory settings (in-laboratory gait speed). Takayanagi et al. 21 demonstrated a weak association (r = 0.333, p < 0.001) between DGS and in-laboratory gait speed; therefore, we cannot simply compare these cut-off points with DGS. However, although a few studies reported cut-off points assessed using DGS, the present study demonstrated that DGS > 1.37 m/s might lead to decreased abdominal and general obesity in elderly people.
The strengths of this study include the relatively large number of samples with a wide range of ages and the objectively measured DGS assessed by a tri-axial accelerometer. Despite these strengths, our study has several limitations. First, the cross-sectional design of our study could not establish a causal relationship between DGS and obesity status. A longitudinal study would be necessary to determine causality and the associated mechanisms. Second, although similar participation rates have been reported in other studies that have used accelerometers 48 , the loss of participants due to insufficient accelerometer data could have resulted in a selection bias in this study. Finally, as this study was confined to participants from a particular country, region, and race, reproducibility should be confirmed by the inclusion of participants from different regions and/or races.

Conclusions
In conclusion, the association between DGS and abdominal and general obesity differed significantly by age. In OA, DGS was significantly and negatively associated with abdominal obesity and general obesity, whereas no significant associations were found in YA. These data could aid in raising awareness about the self-management of obesity via DGS monitoring, especially in case of OA. The effect of age on the relationship between DGS and obesity warrants further investigation.