Ageing Effects on Tripping Risk: The Foot-Ground Clearance of Healthy Community Dwelling Japanese Cohorts Aged 50, 60 and 70 Years

Falls-related injuries are particularly serious for older people, causing pain, reduced community engagement and associated medical costs. Tripping is the leading cause of falls and the current study examined whether minimum ground clearance (MFC) of the swing foot, indicating high tripping risk, would be differentiated across cohorts of healthy 50-, 60- and 70-years old community residents in Japan. Three groups (50s, 60s and 70s) of 123 Konosu City residents consented to be recorded when walking on an unobstructed surface at preferred speed. Gait biomechanics was measured using high speed (100 Hz) motion capture (OptiTrack – Natural Point Inc.), including step length and width, double support, foot contact angle and MFC (swing toe height above the ground). Multivariate Analysis of Variance (MANOVA) was used to conrm ageing effects on MFC and fundamental gait parameters. Pearson's correlations were performed to identify the relationships between mean MFC and other MFC characteristics (SD and SI), step length, step width, double support time and foot contact angle. No other statistical effects were identied for spatio-temporal parameters between the three groups. The 50s cohort MFC was also signicantly higher than 60s and 70s, while step-to-step MFC variability was greater in the 70s than 50s and 60s. Pearson’s correlations demonstrated more symmetrical gait associated with greater MFC height. In the 70s increased MFC height correlated with higher MFC variability and reduced foot contact angle.


Introduction
In demographically ageing societies it is increasingly important to promote healthy ageing to reduce healthcare costs and ensure nancially sustainable social security systems. About one in three senior adults fall at least once a year and up to 20% of cases lead to serious injury or fatality [1]. Previous studies have identi ed tripping as the leading cause of falls [2,3] due to the swing foot striking the walking surface, or an object on it, with su cient force to cause balance loss. Research into ageing effects on walking has, therefore, focussed on identifying foot-ground clearance characteristics that increase tripping risk in older people.
Minimum Foot Clearance (MFC) at mid-swing is critical to determining tripping risk ( Fig. 1) [4] due to the increased probability of surface contact. At this point, the foot's forward velocity also approaches maximum, creating a considerable foot-obstacle contact force. Interestingly, in most previous reports, reduced MFC height has not been reported to be associated with ageing but greater MFC variability, in uencing the MFC distribution, does appear to explain the greater tripping risk as we get older [4].
Analytical approaches to estimating tripping risk have, therefore, modelled the (non-normal) distribution patterns of MFC height and variability [5].
Most previous studies of MFC compared samples of 10 to 30 young adults (18-35 years) with a single group of physically active older people, for example, people over 60 years [6]. For experimental purposes they were, however, undifferentiated with respect to age. The aim of this study was to compare MFC characteristics across three older age groups, comprising individuals in their 50 s, 60 s and 70 s.
Prolonged life expectancy and lower birth rate in Japan are presenting critical social, economic and public health problems, including mobility and falls. The current study's focus was older residents of Konosu City, with a population of 120,000 located in Saitama Prefecture Japan, a predominantly rural region of which more than 98% are Japanese born. More than 30% are over 65 years, higher than the national average (29%). While basic gait assessments, such as the 6 m-walk test [7], are widely used in Japan, there are no reports of ageing effects on gait utilising 3D motion capture. The current research is unique in recruiting a large sample (n = 123) of community dwelling citizens outside the laboratory to obtain precise measures of their walking abilities.
Our study was guided by two research questions. First, what are the effects of chronological age on footground clearance, i.e. MFC central tendency, variability (SD) and left-right asymmetry. Symmetrical control of MFC is affected by ageing [8] and based on previous reports [e.g., 4,8,9] it was hypothesised that MFC height would be lower, more variable and also increasingly asymmetrical across the three age groups. The second question was how ageing effects on stride phase variables in uence MFC height. This was addressed by correlating spatio-temporal gait cycle variables with MFC height across age groups. Alcock et al. [10] found that slower walking was associated with reduced MFC, possibly due to shorter, wider steps and prolonged double support, all of which are seen with advanced age [11]. Flatter foot-ground contact due to reduced dorsi exor strength was hypothesised across age groups; a variable that has also been proposed to correlate with lower foot-ground clearance [12,13].

Apparatus And Procedure
Gait testing was conducted on a 10 m walkway in the City Sports Gymnasium at preferred speed, with trials repeated until a minimum of 30 complete step cycles had been collected. To model foot motion re ective markers were attached to the toe (the superior most distal surface of the foot) and the heel (the most proximal point) [11]. The re ective marker 3D position-time coordinates were sampled at 100 Hz using a three-dimensional (3D) camera system (Optitrack, Natural Point) and then smoothed at 6 Hz using a low-pass Butterworth digital lter. Toe-off and heel contact were identi ed by applying conventional gait event detection algorithms to the heel and toe velocity and acceleration functions [8]. MFC was computed as the toe vertical local minimum within a time-sample window mid-swing, using an in-house algorithm implemented in Visual3D (C-Motion, Inc.) script language [5]. Gait variables were computed from the 3D position-time coordinates of the toe and heel markers with step length and width de ned, respectively, as anterior-posterior and medio-lateral displacements between the heels at heel contact (Fig. 2). Double support time was the period for which both feet were in contact with the walking surface, i.e. from one heel contact to the contralateral toe-off. Foot contact angle was de ned relative to toe and heel positions at heel contact and the walking surface.
As illustrated in Fig. 1, Minimum Foot Clearance (MFC) was the vertical displacement of the toe from the walking surface during mid-swing [4]. All gait data were described using the Mean (central tendency), Standard Deviation (SD) (intra-subject variability) and a Symmetry Index (SI) re ecting left-right differences in the gait variables [8,14]; computed as follows: where R/L indicates right/left foot leading (Fig. 2). The SI was computed for both the mean and SD of each gait variable.

Design And Analysis
A One-way Analysis of Variance (ANOVA) design was used to determine between-group differences in age, height and body mass. Multivariate ANOVA (MANOVA) was used to con rm ageing effects on MFC and fundamental gait parameters. Signi cant overall ANOVA effects were followed up using Tukey's procedure to determine differences between the three age groups. Pearson's correlations were performed to identify the relationships between mean MFC and other MFC characteristics (SD and SI), step length, step width, double support time and foot contact angle. For all test statistics, i.e., F-ratios, Tukey's test, and Pearson's r, signi cant effects were accepted when p-values were less than .05.

Results
One-way ANOVA results con rmed that the three groups were distinguished only by increased age (F 2, 121 = 278.8, p < .01) with no differences in height and body mass.  Fig. 4. No Mean MFC difference was found between the 60 s-70 s groups. In contrast, SD MFC of the 70 s age group was 0.149 cm greater than for the 50 s (p < .05) and 0.106 cm greater than the 60 s cohort (p < .05) but the 50 s-60 s comparison did not reveal a difference in MFC variability. SI MFC was lowest in the 50 s age range but no statistically reliable age effects on SI MFC were con rmed.

Discussion
The sampled population was identi ed as healthy, they were motivated to volunteer for a research project and their fundamental gait parameters were comparable to previous studies with similar age-group samples. Doyo et al. [15], for example, also observed a reduction in step length with age, based on a sample of 2006 community dwelling citizens in Japan. They also reported similar age-related changes in spatio-temporal parameters in the 70 s age range, such as reduced step length.
Two important ndings emerged from the MFC analysis. First, contrary to previous ndings [4,6,8] age effects on MFC height were found because mean MFC height was signi cantly lower in the 60 s and 70 s than the 50 s group. Most previous studies compared young (e.g. 18-35 yrs.) with a single older group (e.g. > 60 yrs.) [6,17] but our analysis from a considerably larger sample, incorporating three sub-groups, revealed that MFC height may reduce prior to the 60 years age range. Inconsistent MFC height has been identi ed as the primary ageing effect on foot trajectory control, leading to higher tripping risk [4]. Our MFC ndings here suggest that MFC height may begin to fall from the 60 s, while signi cantly more variable MFC may appear later, in the 70 s age group.
Pearson's correlations indicated that across all three age groups more symmetrical gait, re ected in a lower SI, was generally associated with elevated MFC height. Previous work [8,11,12,18] suggested that reduced leg strength with ageing leads to higher asymmetry, causing loss of symmetrical gait control and increased tripping risk. In contrast to Alcock et al. [10], we did not nd signi cant correlations between step length and MFC height but they examined both preferred speed and fast walking, revealing increased MFC due to higher velocity, while the current experiment was conducted only at preferred speed. It may, however, be reasonable to suggest that decreased step length associated with ageing-related declines in walking speed [19,20] is causally related to reduced MFC height with ageing.
While lower MFC height was identi ed from the 60 s, correlation analysis revealed that ankle control may decline later, from the 70 s. This age group showed a positive correlation between mean MFC and SD of MFC, such that the positive effects of elevated MFC were counteracted by increased MFC variability [22]. With ageing, the loss of nely coordinated ankle movement may require a greater contribution from the knee and hip, but these joints are less adapted to precise swing foot control [13,23]. Increased Foot Contact Angle was also linked to reduced MFC height only in the 70 s group, also demonstrating impaired ankle action. Heel contact was associated with dorsi exion but correlation results suggested that attempts to achieve increased foot contact angle may have caused reduced MFC in the 70 s participants.
Reduced MFC height was seen from age 60 years while MFC variability increased from 70 years. While each decade showed different strategies to control MFC, in general, less variable and more symmetrical gait optimises MFC control. Exercise interventions may help in maintaining foot elevation and reducing tripping risk and in addition to maintaining ankle dorsi exion, particular at mid-swing close to MFC [13], exercises for older people should emphasise symmetrical walking. Treadmill-based gait training with realtime biofeedback, for example, can increase MFC height while reducing variability [21] and gait-feedback provided by "smart footwear" systems may also reduce tripping risk by alerting the wearer to asymmetrical gait control [25,26].
Precise gait measures obtained using motion capture will more reliably identify age-associated changes to mobility than more commonly used assessments, such as the 6 m-walk test [7,16]. Large-sample community-based gait screening could also be practically undertaken using a force-sensitive commercial gait assessment system (e.g. GaitRite mat) that does not require specialised skills. For a comprehensive understanding of ground clearance, however, 3D analysis is required. Our study used motion capture apparatus, but larger samples could be tested more e ciently using markerless motion capture suits or footwear-mounted wearable sensors. As far as we know, there have been no previous attempts to use 3D motion capture to examine mobility within an everyday community.
This study was conducted as part of Konosu City's health promotion initiative and advances in remote gait monitoring, i.e. gait measurement outside the laboratory, will encourage future falls prevention and physical activity initiatives. This early-stage gait assessment scheme should, therefore, be viewed as a community model with the potential to be adopted by other cities to maintain the mobility and safety of their valuable and deserving senior citizens.

Consent for publication
Not applicable Availability of data and materials The datasets generated and analyses performed during the current study are not publicly available due to the consent requirement of participants, but anonymous descriptive data are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.  Step Length and Step Width (left) and Foot Contact Angle, the angle formed between the toe-heel axis and the walking surface(right). Red dots represent the toe and heel markers. Gait parameter Mean, SD and SI for three age groups. Standard deviation bars indicate intrasubject SD, dots indicate SI shown on the right axis. Figure 4