Walking barefoot vs. with minimalist footwear – influence on gait in younger and older adults

In a randomized within-subject study design, gait data of 31 healthy younger (17 female, 14 male, age: 29 ± 4 years; BMI: 23 ± 2) and 33 healthy community-dwelling older (20 female, 13 male, age: 71 ± 4 years; BMI: 27 ± 4) participants were collected. We advertised in the local newspaper and at a local sport club to recruit the participants of this study.

Inclusion criteria were the age of ≤35 years for the younger group and ≥ 65 years for the older group. Participants had to be able to walk throughout 5 minutes without the need to pause or the use of assisting equipment. Additionally, participants had to be unfamiliar with regular barefoot walking or walking with minimalist shoes. Any self-reported motor-functional impairments that could affect gait performance, such as acute musculoskeletal disorders or neurological diseases, led to exclusion from the study. All subjects provided their written informed consent to their voluntary participation in this study as well as their allowance for publication, which has been approved by the local ethics committee (protocol no. FSV 16/13) and followed the principles of the Helsinki Declaration.

To capture kinematic data, wireless inertial sensors (MTw2, Xsens Technologies B.V., Enschede, The Netherlands, range of measurement of angular velocity: ±1200 deg/s, sampling rate: 100 Hz) were attached to the participants’ right forefeet using tape. The participants were block randomized to the different test condition sequences by using the computer-based program Research Randomizer (https://​www.​randomizer.​org/​). Further, participants were asked to walk at their preferred walking pace back and forth on a 25 m track inside a sports hall with flat ground. The following test conditions were performed in a randomized and balanced order: 1) barefoot walking and 2) walking with minimalist shoes (leguano classic, leguano GmbH, Buchholz, Germany; material: 53% polyamide, 38% lyocell, 7% polypropylene, 2% elastane, Sole: 100% LIFOLIT®). Participants walked for 3 minutes under each condition. The resting time between the test conditions was 5 minutes. Prior to each of the testing conditions, participants walked for approximately 1 minute back and forth the track to familiarize themselves with the corresponding test condition and to improve reliability [24].

The data analysis included gait variability measures as well as the local dynamic gait stability for which the largest Lyapunov exponent (λ) was calculated. We removed the first strides (from the start to the first turn after 25 m) of each trial as well as the last strides (from the last turn to the stop) to avoid possible transients, e.g. to analyze steady-state gait [25]. Furthermore, we removed the first 2.5 m of each section between two successive turns to exclude the acceleration periods after turning [26]. To account for possible deceleration periods before turning, the last 2.5 m were excluded from the subsequent analysis as well. At least 50 strides should be analyzed for gait variability measures [27]. In this study, the first 80 valid strides within each trial were analyzed.

To calculate the gait measures a) stride length, b) stride time and c) minimum toe clearance (MTC), we used a published algorithm [25]. As gait variability measures, the intraindividual standard deviations of each gait measure were analysed. To quantify λ, we time-normalized the three-dimensional angular velocity data of the 80 valid strides to 8000 samples. To reconstruct the state space, we applied the embedding approach. The time delay (9 samples) and the embedded dimension (dE = 6) were determined, using the first minimum mutual information [28] and the global false nearest neighbors method, respectively [29]. The λ was determined using the Rosensteins algorithm [30], which we implemented in MATLAB (version 2016a, TheMathWorks BV, Natrick, USA). The Euclidean distance of each nearest neighbor state space was tracked while the mean of the logarithm of the divergence curve was calculated. λ is defined as the slope (linear fit) through 0–0.5 strides. The implementation was validated previously [25]. Using IBM SPSS Statistics (V 21.0) for all statistical procedures, a two factor (condition: barefoot and minimalist footwear; age: younger and older adults) variance analysis with repeated measures was applied to identify differences across the two conditions.

We want to acknowledge four main study limitations. First, this research involved healthy community-dwelling older adults. Therefore, our current findings may not apply to older adults with health issues or already existing impairments. Most older adults have a higher risk of falling when confronted with existing dysfunctions concerning the gait pattern [34]. This fact points to the potential of future research within clinical settings.

Second, the participants of this study did not walk in their own shoes. Implementing this third condition to the testing procedure could have revealed more knowledge of how minimalist shoes influence gait in comparison to normal shoes, which are worn in daily life. Additionally, it would have made our results even more comparable to other studies. However, this question was not subject of our current study. We also ran our tests for only one type of minimalist shoe. Considering the growth of diversity in products, we encourage future research to compare the influence of different minimalist footwear as well as to include aspects of practicability regarding minimalist shoes for older adults.

Third, the application side in this study was a gym, proving a flat ground. Thus, our findings are limited to overground walking on an even surface. According to Zurales et al. [35], however, uneven surfaces are strong predictors of falls among older adults. Additionally and in line with Li et al. [36], falls occurred more often outdoors than indoors among older adults. Thus, future research should consider testing on an uneven surface, for instance, outdoors.

Last, this study focused on short-term effects. We suggest that future research should target long-term effects of fall prevention training with minimalist shoes or extended periods of wearing minimalist shoes during daily activity. Certainly, a systematic overview of the existing literature would enrich this scope of research.