Introduction

Preferred walking speed (PWS), also known as usual, spontaneous, self-selected or comfortable gait speed, represents a performance measure of mobility in older individuals. PWS, considered as the final and integrated output of the locomotor system, is an indicator of current health and functional status, independent living, quality of life, response to rehabilitation and all causes of mortality in older adults (see for review Middleton et al. 2015) and can be described as the “sixth vital sign” (Fritz and Lusardi 2009). PWS declines with age, with the rate of decline increasing after the critical age of ~65 years (Bohannon and Williams Andrews 2011; Himann et al. 1988). It has been suggested that a gait speed slower than 1 m.s−1 should be considered to indicate a person at high risk of adverse health outcomes, and an improvement or a reduction in PWS of at least 0.1 m.s−1 is a useful predictor for well-being or poor health, respectively, in older people (Fritz and Lusardi 2009). Age is associated with reduced stride length and frequency, anterior pelvic tilt, reduced plantarflexion and reduced propulsive mechanical power generation of the trailing limb during the stance phase with a concomitant distal-to-proximal shift in power generation in older compared to young individuals walking at slower PWS (Franz 2016; Kerrigan et al. 1998) or at the same speed (Boyer et al. 2012; DeVita and Hortobagyi 2000; Hortobagyi et al. 2016). Although PWS decline is age-related and its underlying factors have not yet been completely elucidated, changes in muscle strength (Bassey et al. 1988; Buchner et al. 1996), balance (Murray et al. 1969), aerobic fitness (Malatesta et al. 2004) and depressive symptoms (Buchner et al. 1996) have been shown to be important determinants of PWS in older individuals. Moreover, increased physical activity level can maintain PWS and mitigate declines in the mechanics of walking with age (Boyer et al. 2012; Savelberg et al. 2007). Compared to young adults, physically active older individuals did not exhibit reduced PWS, even if the latter maintained PWS by increasing stride frequency and decreasing stride length and by a more plantar-flexed ankle at heel-strike and a less plantar-flexed ankle at the toe-off (Boyer et al. 2012). However, these changes were smaller than those found in sedentary older adults (Boyer et al. 2012; Himann et al. 1988; Riley et al. 2001).

PWS is usually assessed during overground (via a 2–40 m walkway) (see for review Middleton et al. 2015) or treadmill walking (Dal et al. 2010; Malatesta et al. 2003, 2004; Martin et al. 1992; Nagano et al. 2013) in older individuals. There is a controversy regarding the similarity between treadmill and overground walking in gait analysis. During treadmill walking, a recalibration of the motor output and a higher “neuronal-computational” effort seem necessary because of the visual–kinaesthetic conflict (Zanetti and Schieppati 2007). However, it has been reported that similar neural networks may be involved in treadmill and overground walking (Choi and Bastian 2007) and that there are only minor kinetic and spatiotemporal changes between these gait modalities in healthy individuals (Lee and Hidler 2008; Warabi et al. 2005). Only one study compared PWS determination and its spatiotemporal parameters using treadmill and overground assessments in young and older adults (Nagano et al. 2013). The findings of this study showed that PWS was similar for overground walking between the two age groups, but only older adults walked at a slower PWS with reduced step length, increased step frequency and double support duration during treadmill compared to overground walking. These mechanical changes may indicate that treadmill walking is more destabilizing and induces “safety-related” gait adaptations (Nagano et al. 2013). However, these mechanical changes may also be due to incomplete familiarization with the treadmill in the older adults, who performed only 10 min of acclimation to the treadmill in this study. In fact, this duration has been shown to be insufficient to properly familiarize older individuals to the use of a treadmill (Wass et al. 2005). Moreover, Nagano et al. (2013) tested the same two age groups only at PWS under the two walking conditions (overground vs. treadmill) and not at a fixed speed. Therefore, it is difficult to differentiate the effect of the speed or walking conditions on the spatiotemporal parameters of walking in the young and older individuals in this study, whose physical activity level was not directly assessed.

Therefore, the aim of this study was twofold: (1) to compare the effect of treadmill and overground walking on PWS, spatiotemporal parameters and foot kinematics in healthy active older and young adults after adequate treadmill familiarization for both groups; and (2) to compare the spatiotemporal parameters and foot kinematics of the two groups walking at the same speed (i.e., overground PWS on the treadmill) under the two walking conditions. We hypothesized that PWS would be similar in the two groups during treadmill and overground walking and that no differences would be found between the two walking modalities. Moreover, older adults would display an increased stride frequency, a decreased stride length and a more plantar-flexed ankle at heel-strike and a less plantar-flexed ankle at the toe-off under the two walking conditions compared to young adults.

Methods

Participants

Twenty-five older (O; 11 women and 14 men; 72.2 ± 4.0, range 66–80 years) and 20 young (Y; 10 women and 10 men; 24.4 ± 2.1, range 20–30 years) adults participated in the study. The older adults (aged ≥65 years) were recruited from associations that offer various physical activities for older individuals, including regular country walks and gymnastics. The participants were healthy and independent and participated in these physical activity programs at least once each week for 9 months of each year. All subjects were medically healthy and free of clinically significant orthopedic, neurological, cardiovascular or respiratory problems. The young participants were sport science students, healthy and physically active. The protocol and the consent form were approved by the local ethics committee, and all participants provided informed written consent.

Experimental design

Each participant completed one test session, including three parts with 5 min of recovery between each part. In the first part of the test session, the subjects were introduced to the experimental procedure, and their anthropometric characteristics and physical activity level were assessed. In the second part of the test session, PWS and spatiotemporal parameters at this speed were evaluated during overground walking. In the third part of the test session, each participant was familiarized with treadmill walking: 10 and 30 min of treadmill accommodation across different walking speeds (0.56, 0.83, 1.11, 1.39 and 1.67 m·s−1) were used for young (Van de Putte et al. 2006; Wall and Charteris 1981) and older (Malatesta et al. 2003) individuals, respectively. After a brief rest period, PWS was determined in a randomized order according to procedures proposed by (1) Martin et al. (1992) and (2) Nagano et al. (2013) (procedure adapted). The spatiotemporal parameters and foot kinematics were assessed at these speeds and at the PWS determined using overground walking (PWSO,T) during 3 min of treadmill walking at each speed with at least 5 min of recovery between each trial. PWS was then determined a second time according to the procedures proposed by Martin et al. (1992). For each walking trial, ten consecutive strides were collected and selected for biomechanical analysis.

Assessments

Anthropometric characteristics

Standing height was measured using a Harpenden stadiometer. Body mass was measured to the nearest 0.1 kg using a precision digital scale with the subject wearing shorts and a T-shirt. Lower limb length was assessed in the standing position as the distance between the great trochanter and the ground for the right lower limb.

Physical activity level

The physical activity of the older participants was estimated from a physical activity questionnaire that was positively correlated with repeated 24-h activity recalls, pedometer measurements and test–retest reliability in this population (Voorrips et al. 1991). Different scores were used to quantify household activities, sports activities and other leisure time activities involving physical activity. The questionnaire provides a method for classifying older individuals into categories of high, medium and low physical activity, with cut-off points of 9.4 and 16.5 for total scores. Each young participant completed a self-reported measurement of habitual physical activity questionnaire divided into three sections: physical activity at work, sport during leisure time, and physical activity during leisure excluding sport (Baecke et al. 1982).

Overground preferred walking speed

The participants were asked to walk on a 25-m walkway at their PWS (PWSO) wearing two inertial sensor system Physilogs® (GaitUp, Switzerland) integrating a micro-controller, memory, 3-axis accelerometer (range ±3 g), gyroscope (range ±800° s−1) and battery comprising a small (50 mm × 37 mm × 9.2 mm) and lightweight (19 g) module attached to the shoes with an elastic Velcro strap. These sensors were used to measure the speed, spatiotemporal parameters and foot kinematics of walking and were previously validated to assess gait pattern in young (Mariani et al. 2010) and older adults (Dadashi et al. 2014; Mariani et al. 2010). Compared to an optical motion capture system, accuracy and precision in the assessment of walking speed, spatiotemporal parameters and foot kinematics of the two inertial sensor system Physilogs® were good, and showed an excellent repeatability across measurements (test–retest reliability) (Mariani et al. 2010). The participants performed three trials on a 25-m walkway with a brief recovery between trials; the biomechanical assessments were measured continuously during each walking trial, but only the ten strides performed in the middle of the 25-m bout of walking were selected and analyzed to limit the effect of acceleration and deceleration phases at the beginning and end of the walkway.

Treadmill preferred walking speeds

PWS was assessed in a randomized order using two procedures; one proposed by Martin et al. (1992) and the other adapted from Nagano et al. (2013). In the former, the participants started to walk on the treadmill (T150–FMT–MED, Arsalis, Belgium) at 0.56 m·s−1 without receiving any feedback regarding their speed. The speed was then slowly increased until the individual subjectively identified his PWS, and this speed was maintained for 1 min and slightly modified according to the participant’s directions. This procedure was repeated starting from the highest familiarization speed (1.67 m·s−1) or from the previously determined PWS of +0.42 m·s−1 (when PWS >1.25 m·s−1) and then gradually reducing the speed to the individual subjective PWS. The average of the two speeds selected by the participant (i.e., during the trials in which the speed was increased and decreased) was considered the final PWS (PWST,1). In the second procedure used, the participants started to walk at PWSO for 2 min without receiving any feedback regarding their speed; this speed was then slightly modified according to the participant’s directions to determine their PWS. Once this speed was selected, it was decreased and increased by 0.3 m·s−1 to determine the actual PWS of each subject (PWST,2). After each determination of the two treadmill PWS measurements, the participants walked at these speeds for 3 min, and spatiotemporal parameters were assessed continuously using two inertial sensor system Physilogs®; however, only ten strides of the last 30 s were selected and analyzed. At the end of the experimental test session, the PWS assessment procedure of Martin et al. (1992) was performed a second time to examine the test–retest reliability of this protocol (PWST,1bis).

Spatiotemporal parameters and foot kinematics

Using the two inertial sensor system Physilogs®, with dedicated algorithms, and based on the detection of temporal parameters, coupled to optimized fusion and de-drifted integration of inertial signals, the speed, spatiotemporal parameters and foot kinematics of walking were assessed for each walking trial as previously described (Dadashi et al. 2014; Mariani et al. 2010). The measured temporal parameters were stride duration and frequency as well as relative duration of swing, stance and double support (percent of gait cycle duration). The relative duration of the three inner-stance phases comprising the stance (loading response, foot-flat and push-off) were computed and expressed as percentages of gait cycle duration. The measured spatial parameters were stride length, swing width. The foot kinematics were heel-strike pitch angle (the positive angle formed by the ground and the longitudinal axis of the foot during heel-strike), toe-off pitch angle (the negative angle formed by the ground and the longitudinal axis of the foot during toe-off) and foot clearance. The latter was defined as the foot’s height during the swing phase and was characterized by maximal heel clearance, maximal toe clearance (first and second maximal) and minimal toe clearance.

Rating of perceived effort (RPE)

The 6–20 Borg scale (Borg 1982) was used to evaluate perceived exertion after each walking trial for older participants only.

Statistical analysis

Data are expressed as the mean ± SD for all variables. A t test was used to test the differences between the anthropometric characteristics of the two groups. A two-way repeated-measures mixed-design ANOVA [speed (PWSO vs. PWST,1 vs. PWST,2) × group (older vs. young)] followed by contrasts was used to test the difference between the different PWSs. For comparing PWSO, PWST,1 and PWSO,T, a two-way repeated-measures mixed-design ANOVA [speed (PWSO vs. PWST,1 vs. PWSO,T) × group (older vs. young)] followed by contrasts was used to determine the effects of the three walking speeds on spatiotemporal parameters and foot kinematics in the two groups. A one-way repeated-measures ANOVA was performed to determine the effects of the walking trails on RPE in older participants. To compare the agreement between the assessments of PWS using the protocol proposed by Martin et al. (1992) (PWST,1 and PWST,1bis), the Pearson product-moment correlation coefficient was calculated, and Bland–Altman plots (Bland and Altman 1986) were graphed. The level of significance was set at P ≤ 0.05.

Results

Participants

The anthropometric characteristics of the study participants are presented in Table 1. There was no significant difference between the two groups in terms of body mass (P = 0.052) and lower limb length (P = 0.51). Height was significantly greater (P = 0.03) and BMI was significantly lower (P < 0.001) in young individuals than in older individuals. In the older group, the mean physical activity level estimated from Voorrips et al. (1991) questionnaire was 14.1 ± 5.7 (low physical activity level: n = 5; moderate physical activity level: n = 14; high physical activity level: n = 6). In the young group, the mean physical activity level estimated from Baecke et al. (1982) questionnaire was 9.1 ± 1.5, confirming that our young participants were physically active (Cheneviere et al. 2009).

Table 1 Anthropometric characteristics
Full size table

Preferred walking speed

PWSO was significantly faster compared with PWST,1 and PWST,2 (P < 0.001; Fig. 1), and there was no significant difference between the groups (group effect: P = 0.16). There was no significant difference between the two treadmill PWS values (P = 0.86; Fig. 1). Moreover, PWST,1 and PWST,2 were significantly correlated (r = 0.92; P < 0.001). For these reasons and for the sake of clarity, PWST,1 will be used to represent treadmill PWS throughout the manuscript. Considering both groups together, PWSO and PWST,1 were significantly correlated (r = 0.68; P < 0.001). Moreover, there was no significant difference between PWST,1 (1.31 ± 0.17) and PWST,1bis (1.32 ± 0.18; P = 0.11), and these values were significantly and strongly correlated and were close to the line of identity (r = 0.97; P < 0.001; Fig. 2). This was confirmed by our finding that the biases were close to zero (−0.01 ± 0.05 m·s−1) and that the limits of agreement were narrow (from −0.10 to 0.08 m·s−1; Fig. 2).

Fig. 1
figure 1

Preferred walking speed selected by older and young participants during overground (PWSO) and treadmill walking using the protocols proposed by Martin et al. (1992) (PWST,1) and adapted from Nagano et al. (2013) (PWST,2). Asterisk significant difference from PWSO (P < 0.005)

Full size image
Fig. 2
figure 2

Correlation (a) and Bland–Altman plot graphing (b) between the preferred walking speeds (PWST,1 and PWST,1bis) as determined on a treadmill using the protocol of Martin et al. (1992). In a, the dotted line represents the line of identity. In b, the light dotted line represents the bias from the mean difference, and the dark dotted line represents the upper and lower 95% limits of agreement

Full size image

Spatiotemporal parameters and foot kinematics

Temporal parameters

A significant speed effect was found for all temporal parameters (P ≤ 0.04; Table 2). Stride duration was significantly shorter but stride frequency was significantly higher for PWSO and PWSO,T when compared with PWST,1 for both groups (P < 0.001 for both; Table 2). Moreover, only these two variables showed a significant group effect (P ≤ 0.001), and stride duration was significantly shorter but stride frequency was significantly higher in older individuals than in young individuals. Relative loading response duration was significantly higher but relative foot-flat duration was significantly lower for PWSO and PWSO,T than for PWST,1 for both groups (P < 0.001 for both; Table 2). Relative stance and double support durations were significantly longer but swing duration was significantly shorter for treadmill walking compared with overground walking in both groups (P ≤ 0.04; Table 2). Relative push-off duration was significantly lower for PWSO and PWST,1 when compared with PWSO,T (P = 0.003; Table 2). Under the two walking conditions, relative push-off duration was significantly correlated with the respective PWS (overground: r = 0.65; P < 0.001; treadmill: r = 0.73; P < 0.001).

Table 2 Spatiotemporal parameters and foot kinematics during overground and treadmill walking
Full size table

Spatial parameters

A significant speed effect was found for all spatial parameters (P ≤ 0.001; Table 2). At PWSO during treadmill and overground walking, stride length was significantly higher compared to that for PWST,1 in both groups (P < 0.001; Table 2). Swing width showed a significant interaction effect (Table 2); significantly higher values were found for PWSO than for PWST,1 and PWSO,T in young individuals (P ≤ 0.04), and significantly lower values were found for PWST,1 compared to PWSO,T in older participants (P = 0.01).

Foot kinematics

At PWSO during treadmill and overground walking, toe-off pitch angle was significantly higher compared to those for PWST,1 in both groups (P < 0.001; Table 2). Toe-off pitch angle was significantly higher in older individuals than in young individuals (group effect: P = 0.013). Heel-strike pitch angle was significantly higher for PWSO than for PWST,1 and PWSO,T (P < 0.001) and for PWST,1 than for PWSO,T in the two groups (P = 0.005; Table 2). Maximal angular velocity during swing was significantly slower during treadmill walking at PWST,1 (406.8 ± 44.4 °.s−1) compared with treadmill and overground walking at PWSO (426.7 ± 44.1 and 422.0 ± 44.3 and °.s−1, respectively; P ≤ 0.001) in both groups, and significantly higher values were found in older individuals (429.6 ± 46.6 and °.s−1) than in young individuals (405.0° ± 38.5 °.s−1; P = 0.049). A significant speed effect was found only for the second maximal and minimal toe clearance (P ≤ 0.03; Table 2). The latter was significantly higher for PWSO than for PWSO,T in both groups (P < 0.001), and significantly lower values were found in young individuals than in older individuals (P = 0.01; Table 2). The second maximal toe clearance was significantly lower for PWST,1 and PWSO,T than for PWSO in the two groups (P = 0.03; Table 2). No significant difference was found for the first maximal toe clearance in both groups (a significant interaction effect with no significant differences was found with the contrasts; Table 2). Maximal heel clearance showed a significant interaction effect; significantly lower values were found for PWSO compared with PWST,1 in older participants only (P = 0.03; Table 2).

Rating of perceived effort

In older participants, RPE at PWSO (6.8 ± 1.4) was significantly lower than RPE at PWST,1 (9.7 ± 2.2), PWST,2 (9.8 ± 2.2) and PWSO,T (10.1 ± 2.3; P < 0.001), and no significant difference was found among the three treadmill walking trials (P > 0.21).

Discussion

This study showed that healthy and physically active older adults who were adequately familiarized to the treadmill selected a PWS similar to that selected by young individuals during overground and treadmill walking with only attenuated spatiotemporal and foot kinematic changes (greater stride frequency and min toe clearance and a less plantar-flexed ankle at the toe-off). However, in contrast with our hypothesis, the two groups walked at a slower PWS on the treadmill than during overground walking, and “safety-related” (Nagano et al. 2013) gait kinematic adaptations were small. Moreover, the participants in both groups selected a PWS that minimized the push-off duration during the two walking conditions.

Preferred walking speed

The PWSO and PWST values of the two groups are consistent with previous results (Bohannon and Williams Andrews 2011; Boyer et al. 2012; Himann et al. 1988; Malatesta et al. 2003; Martin et al. 1992; Nagano et al. 2013) and corroborate that our older adults were physically active (Boyer et al. 2012). In fact, the mean value of PWSO in the older group (1.45 m.s−1) was faster than the cut-off value (1.3 m.s−1) reported for “extremely fit” older adults (Ainsworth et al. 2011; Middleton et al. 2015; Studenski 2009).

Under the two walking conditions, PWS was similar between the two groups, as previously shown for PWSO (Boyer et al. 2012) and PWST (Martin et al. 1992). This shows that physically active older individuals can maintain PWS as assessed during overground or treadmill walking and mitigate its aging-related decline.

The two groups selected a faster PWS during overground walking compared with treadmill walking (+15% for older individuals and +9% for young individuals). These findings only partially confirm those of Nagano et al. (2013), who reported that PWSO was 15% faster than PWST in older individuals only but that the two PWS values were similar in young individuals. However, our results for the young group are consistent with those obtained by Dal et al. (2010), who used a similar treadmill protocol (independent of PWSO) to determine PWS, while the protocol of Nagano et al. (2013) began at the previously assessed PWSO, likely influencing the selection of PWST in young individuals. Globally, our findings suggest that, even though the physically active individuals were familiarized with the treadmill, walking on a treadmill requires greater balance that may lead to slower PWST compared to PWSO regardless of age (Dal et al. 2010; Nagano et al. 2013). This was indirectly confirmed by the higher RPE measured during treadmill walking compared to overground walking at the PWS or at a fixed speed in the older group of this study and as previously reported (Marsh et al. 2006). However, for both groups, the two PWS values are in the region of the graph where the energy cost–walking speed relationship is relatively flat (i.e., the bottom of the U-shaped curve) and are close to the optimal walking speed (Mian et al. 2006). Therefore, treadmill walking can be used to determine PWS during the functional screening of older adults; however, there could be a slight underestimation of actual PWSO. In addition, PWST as determined using the protocol of Martin et al. (1992), which is independent of PWSO and takes hysteresis into account, was highly reproducible (Fig. 2). This confirms that this protocol can be used to assess PWS before and after an intervention to test its efficacy in older adults (Malatesta et al. 2010).

Spatiotemporal parameters and foot kinematics

Older vs. young individuals

A main age (group) effect was found for stride frequency and duration, toe-off pitch angle and min toe clearance (Tables 2, 3). Stride frequency was ~8% higher in older individuals than in young individuals while walking at non-significant different speeds independently of gait conditions. This confirms previous results suggesting that increased levels of physical activity can maintain PWS with attenuated changes in gait mechanics (Boyer et al. 2012). Toe-off pitch angle was ~5% lower, whereas minimum foot height (i.e., min toe clearance) and maximal angular velocity during the swing phase were ~31 and ~6% higher, respectively, in older individuals compared with young adults. The former compensated for the lower plantarflexion of the ankle and elevation of the heel during the toe-off, most likely due to the lower plantarflexor power during the late stance phase compared with young individuals (Judge et al. 1996), by elevating and speeding the foot more during the swing phase. These findings indirectly corroborate that older adults, walking at higher stride frequency and lower toe-off pitch angle, may compensate for reductions in plantarflexor power by increasing hip flexor power (Boyer et al. 2012; DeVita and Hortobagyi 2000; Judge et al. 1996).

Table 3 Schematic representation of the different effects on the spatiotemporal parameters and foot kinematics of walking in older and young participants
Full size table

Overground vs. treadmill walking

The comparison of PWST,1 and PWSO measured on the treadmill (PWSO,T) with PWSO in both groups allows us to isolate the effect of the walking conditions independently of variations in speed (Tables 2, 3). In both groups, our findings showed that treadmill walking was characterized by increased stance, double support duration and max heel clearance and reduced swing duration, heel-strike pitch angle, second max and min toe clearance. Independently of age, the participants walked on the treadmill using a “sliding” gait pattern, thus maximizing the stance phase duration and confirming previous findings in older adults with shorter treadmill familiarization (Nagano et al. 2013; Watt et al. 2010) and showing the presence of “safety-related” gait adaptations (Nagano et al. 2013) in treadmill walking. Moreover, in the same direction as these adaptations, swing width was decreased on the treadmill in older adults only. However, as previously reported in young individuals (Riley et al. 2007; Tesio and Rota 2008), the magnitude of these changes was relatively small, corroborating that treadmill and overground walking are very similar and that the former can be used to assess gait kinematics in independent and physically active older adults who have been adequately familiarized with walking on a treadmill. Moreover, on treadmill, a greater number of strides at constant speed can be sampled and analyzed, giving more accurate results (Owings and Grabiner 2003; Pavei et al. 2017).

Push-off

In both groups and gait conditions, the relative push-off duration (~35% of gait cycle duration) was significantly shorter during walking at the PWS than at an imposed and fixed walking speed on the treadmill (PWSO,T was faster than PWST,1) and was significantly correlated with the respective PWS. These findings may suggest that participants selected their PWS while minimizing push-off duration during treadmill and overground walking. This may allow individuals to improve contractile conditions (i.e., by developing a favorable shortening velocity and an enhanced use of tendinous tissue elasticity) during force generation in the plantarflexor muscles to optimize the production of mechanical work and decrease the energy cost of walking at the PWS under both conditions as previously reported in older adults walking at PWSO (Stenroth et al. 2016). These authors suggested that such a speed-related adaptation in plantarflexion muscle–tendon function could be pivotal for the selection of PWS in this population. Moreover, limiting the propulsion during push-off may improve balance in older adults, confirming that the selection of PWS may be multifactorial and result from competing factors and constraints (Franz and Kram 2014; Stenroth et al. 2016).

Methodological limitations

Our healthy and physically active older adults may be not representative of the general older population. However, this choice in the enrollment was necessary to study the effect of a high physical activity level and “normal aging” on the selection of PWS during overground and treadmill walking.

Conclusions

This study showed that maintaining a high level of physical activity allows older adults to preserve PWS, as assessed during overground or treadmill walking, and mitigate its aging-related decline with attenuated spatiotemporal and foot kinematic changes compared to those characterizing sedentary older adults. However, the two groups walked at a slower PWS on the treadmill than during overground walking due to the use of a “sliding” gait pattern, thus maximizing the stance but adding only small “safety-related” (Nagano et al. 2013) kinematic gait adaptations. This suggests that treadmill and overground walking are very similar and that the former can be used for assessing gait kinematics in independent and physically active older adults who are adequately familiarized to the treadmill. In fact, independently of walking conditions, minimizing the push-off duration seemed to play a pivotal role in the selection of PWS in both groups. However, future studies should examine the neuromuscular and mechanical factors that are involved in minimizing the push-off duration in the two studied walking conditions.