- Email: [email protected]
Foot motion in children shoes—A comparison of barefoot walking with shod walking in conventional and flexible shoes
Gait & Posture 27 (2008) 51–59 www.elsevier.com/locate/gaitpost
Foot motion in children shoes—A comparison of barefoot walking with shod walking in conventional and flexible shoes Sebastian Wolf a,*, Jan Simon b, Dimitrios Patikas a, Waltraud Schuster a, Petra Armbrust a, Leonhard Do¨derlein a b
a Department of Orthopedic Surgery, University of Heidelberg, Schlierbacher Landstrasse 200a, 69118 Heidelberg, Germany Interdisciplinary Center for Scientific Computation, Heidelberg (IWR), Im Neuenheimer Feld 368, 69120 Heidelberg, Germany
Received 13 July 2006; received in revised form 11 January 2007; accepted 13 January 2007
Abstract The increased prevalence for flatfoot and hallux valgus in modern societies may be the consequence of inaequate footwear in childhood. Based on the assumption that barefoot walking represents the best condition for the development of a healthy foot the objective of this study was to monitor the influence of commercial footwear on children’s foot motion during walking. Furthermore, an attempt was made to reduce this influence by changing the physical properties of standard footwear. Children’s barefoot motion pattern was monitored by a marker-based optical 3D-tracking method using a multi-segment foot model. In the study’s first stage, barefoot walking was compared to walking with a commercial product. In the second stage it was compared to both, the pattern with the commercial product and with the shoe modified on the basis of the findings of the first stage of the study. Eighteen children (8.2 0.7 years old) with no foot deformity and with the same shoe size were recruited for this study. It was found that tibio-talar ROM increased in the commercial shoe (26.68) compared to the barefoot condition (22.58, p = 0.001) whereas the medial arch changes for push-off were diminished since the variation in arch length was reduced from 9.9% (barefoot) to 5.9% (shoe, p < 0.001). Further, ROM in foot torsion along the long foot axis was reduced from 9.88 (bare) to 4.78 (shoe, p < 0.001). These parameters could be improved with more flexible footwear. The present study shows that slimmer and more flexible children’s shoes do not change foot motion as much as conventional shoes and therefore should be recommended not only for children in this age but for healthy children in general. # 2007 Elsevier B.V. All rights reserved. Keywords: Footwear; Children; Motion analysis
1. Introduction Shoes are primarily used to protect the foot from injuries due to rough or uneven ground surfaces and from excessive impact due to a hard ground. Furthermore, shoes protect the feet from a cold and wet environment. This holds true, in principle, for children’s shoes as well. However, it can be claimed, that optimum foot development can only occur in barefoot conditions. Stiff and tight footwear may lead to deformity and stiffness [1]. The design of children’s shoes should be based on the barefoot model taking into account shock absorption and load distribution [1]. There are several * Corresponding author. Tel.: +49 6221 966724; fax: +49 6221 966725. E-mail address: [email protected] (S. Wolf). 0966-6362/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2007.01.005
studies indicating that inadequate footwear or even footwear in general may affect the physiological development of the foot. Rao and Joseph [2] showed that in 2300 children between the age of 4 and 13 years the incidence of a flat foot among those who used footwear (8.6%) was significantly higher compared to those who did not (2.8%). This effect was found in all age groups. Moreover, children wearing closed shoes showed a higher incidence for flat feet than those wearing sandals or slippers. Jerosch and Mamsch [3] documented that in children between 10 and 13 years of age only 36.5% had normal feet. The rest had mild to significant deformities, usually valgus (39.4%). Flat foot deformity was diagnosed in 19.1% and a hallux valgus in 17.1% of the children. Little is known about the direct effects of foot wear on the kinematic behaviour of a child’s foot. Fluoroscopic
52
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
measurements are possible [4], but ethically inappropriate in children. Optical 3D tracking of markers placed on the foot is a valuable tool for the analysis of foot kinematics [5–10] which is feasible also for children [11] but, to our knowledge, has never been applied to shod walking in children. The aim of this study was to apply a marker-based optical method to bare foot walking as well as shod walking by perforating the shoe at locations at which markers were to be applied directly on the foot. The marker protocol used in the Heidelberg Foot Measurement Method (HFMM) [10] has therefore been reduced and the model simplified to monitor the child’s foot movement ‘‘barefoot’’ within the shoe. The aim of the first stage of this prospective study was to assess the influence of a typical and commercially available children’s shoe (shoe1) on foot motion. The second stage aimed at developing a modified shoe (shoe2) based on the findings of stage one as well as the recommendations by Staheli [1]: his study suggested that a thinner and more flexible sole and upper would allow better afferent stimulus as well as better mobility of foot segments. In order to assess the new shoe design, the foot motion patterns in shoe1 and shoe2 were compared to the barefoot pattern.
2. Participants and methods 2.1. Data collection In the recruitment process, more than 100 children between 6 and 10 years of age were initially assessed. Eighteen children (age 8.2 0.7 years, 8 girls, 10 boys) with no foot deformities, previous significant injuries or operations to their feet were elected for the study. All children were fitted with shoes of French size 33. Foot size effects in this small distribution of foot lengths (20.7 0.4 cm) could be eliminated. All children and parents/carers consented according to the local ethics committee regulations. Ten reflective markers of 14 mm diameter were placed on the left shank and foot as shown in Figs. 1 and 2 for the foot kinematic measurements. This marker setup consists of a subset of markers used for the HFMM [10] excluding three markers at the mid-foot. In the barefoot trials, the complete marker set of the HFMM has been used for validation reasons.
Table 1 Physical properties of the shoes Property
Shoe1
Shoe2
Weight (g) Total length (cm) Inner length (cm) Forefoot outer width (cm) Forefoot inner width (cm) Sole thickness heel (mm) Sole thickness forefoot (mm) Profile thickness
245 23.5 21.9 8.9 7.5 23 14 3.0
158 23.2 21.9 8.6 7.5 7 7 1.5
Shoe1: commercial shoe, shoe2: experimental shoe.
Shoe1, a commercial shoe (Elefanten, Germany; Fig. 1, right) with Balmoral closure style was prepared with four additional cut-outs for placing markers at first and fifth metatarsal and the medial and lateral sides of the calcaneus. The markers were placed at the same spot onto the skin as in the barefoot measurements except for the hallux marker which was placed onto the shoe. Without further manipulation, the shoe left enough room to place the malleoli markers onto the skin. Heel and toe markers on both feet were used for event detection. The experimental shoe2, prepared in a similar manner, had a slimmer outer sole, a more flexible upper and was overall lighter (Fig. 3). The insole was identical. Some physical properties are listed in Table 1. In the beginning of each test session, a static trial with the barefoot marker placement was captured in standing position as reference. The children were then asked to walk on a 7 m walkway at self-selected speed monitored by a motion capture system (6 cameras VICON 612 @ 120 Hz). The measurement was repeated with shoe1 and shoe2. The positions of the markers were marked with a felt tip pen on the skin to preserve identical marker positioning after wearing or changing shoes. If the prepared cut-outs in the shoe did not allow an exact placement of a marker, the cutouts were slightly enlarged without compromising the stability of the shoe. Care was taken to avoid any contact between the markers and the rim of the cut-outs. 2.2. Calculation of foot motion parameters In order to measure foot motion, four additional virtual positions were calculated to represent anatomical reference points in the shank and foot: (1) a center point in the calcaneus as midpoint between the two calcaneus markers,
Fig. 1. Marker placement barefoot (left) and in the shoe1 (right). Four cut-outs were used for the medial and lateral calcaneal and metatarsal markers.
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
53
Fig. 2. Markers used for the measurement—LEP, MEP: lateral and medial femoral epicondyle; LML, MML: lateral and medial malleolus; LCL, MCL: lateral, medial calcaneus; MT1, MT5: first (medial) and fifth (lateral) metatarsal head; HLX: distal phalanx of hallux. CCL: heel marker for event detection. Additional virtual points and lines for the model calculation of joint centers, axes and segment lines are indicated.
(2) a forefoot center point defined as a weighted mean position between first (60%) and fifth (40%) metatarsal head marker, (3) the ankle center as the midpoint between the malleoli and (4) the knee center as the midpoint between medial and lateral femoral epicondyles (Fig. 2). Connecting lines between these four reference points in the sagittal plane represent the long axis of the tibia and three axes within the foot: hindfoot axis, midfoot axis and long foot axis (Fig. 2,
right). The connecting line between the physical markers on the malleoli defined the ankle joint axis; metatarsal and calcaneus markers defined a forefoot axis and a calcaneus line, respectively (Table 2). Rotation angles within the foot were defined between pairs of line-like 2D-segments represented by these connecting lines. For simplicity, the same lines were also used to define rotation axes articulating the 2D-segments. The hindfoot axis for example represents the orientation of the calcaneus relative to the long axis of the tibia describing ankle flexion (tibio-talar flexion, Table 3) about the ankle axis i.e. the tibio-talar joint. Further, the hindfoot axis served to define a rotation axis for subtalar joint motion between hindfoot and tibia (Table 3) roughly coinciding with the biomechanical subtalar joint axis. Technically, rotation between pairs of 2D-segments was determined as projection angle by projecting both segments into a plane perpendicular to the rotation axis as previously Table 2 Lines and axes defined by locations of real and virtual marker positions
Fig. 3. Sole of experimental shoe2 (top) and lateral view (bottom).
No.
Line/axis
Marker
1 2 3 4 5 6 7 8 9 10 11
Knee axis Ankle axis Tibia axis Long foot axis Midfoot axis Hind foot axis Calcaneus line Medial arch line Forefoot axis Hallux line Gait direction
MEP–LEP MML–LML Knee center–ankle center Calcaneus center–forefoot center Ankle center–forefoot center Ankle center–calcaneus center MCL–LCL MCL–MT1 MT1–MT5 MT1–HLX CCL (between heel strikes)
54
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
Table 3 Definition of angles between pairs of segments No.
Angle
Axis of rotation
Segment 1
Segment 2
1 2 3 4 5 6 7 8
Tibia-foot flexion Tibio-talar flexion Subtalar rotation Hallux flexion Foot rotation Foot torsion Forefoot supination Foot progression
Ankle axis Ankle axis Hind foot axis Forefoot axis Tibia axis Long foot axis Midfoot axis Tibia axis
Tibia axis Tibia axis Calcaneus line Midfoot axis Long foot axis Forefoot axis Forefoot axis Long foot axis
Long foot axis Hind foot axis Ankle axis Hallux line Knee axis Hind foot axis Ankle axis Gait direction
reported [10]. This was calculated as the scalar product of two vector products e.g.
Furthermore, the time–distance parameters were determined for each subject and condition.
nankle-axis ~ nhindfoot-axis Þ; cosðasubtalar-rotation Þ ¼ hð~ ð~ ncalcaneus-line ~ nhindfoot-axis Þi:
2.3. Statistics (1)
The vectors in Eq. (1) are normalized direction vectors. In the static trial, the anatomic neutral position was defined for all angles. Tibia-foot flexion (no. 1 in Table 3) differed and was calculated directly in the plane defined by the lines of the tibia and the long foot axis which did not perfectly coincide with the normal plane of the ankle axis. The numerical results however were very similar. For further analysis, the unfiltered data of the foot kinematics were time-normalized to the gait cycle and averaged across 7–10 strides for each child and measurement condition. Seven angles describing the orientation between different foot segments were calculated according to Table 3: (1) tibia-foot flexion, in conventional gait analysis referred to as ‘‘ankle flexion’’ or ‘‘dorsiflexion’’ which measures sagittal plane motion of the foot (regarded as one rigid segment) relative to the tibia, (2) tibio-talar flexion, which measures the same kind of movement but refers to markers on tibia and hindfoot only such that mid and forefoot motion is excluded, (3) subtalar rotation, measuring hindfoot motion relative to the tibia about an axis which is close to the functional axis of the subtalar joint, (4) hallux flexion, which addresses sagittal plane motion of the first toe, (5) foot rotation measuring motion of the complete foot relative to the tibia similar to tibia-foot flexion but in the transverse plane, (6) foot torsion, a rather technical parameter measuring transverse plane forefoot motion relative to the hindfoot, and (7) forefoot supination, measuring forefoot motion relative to the ankle. Further, foot progression (8), the orientation of the long foot axis with respect to the gait direction is determined. For validation, correlations with corresponding parameters of the HFMM were calculated. The distance between the first metatarsal marker (MT1) and the medial calcaneus marker (MCL) representing the length of the medial arch was also measured. The relative change to this distance during the static trial was calculated in %. Similarly, the change in forefoot width was estimated by calculating the distance between the markers of first and fifth metatarsal (MT1 and MT5), respectively (Table 3).
For quantification of differences between the measurement conditions, the individual range of motion (ROM) for each angular parameter was determined. The change in medial arch length and forefoot width within the gait cycle (given in %) was calculated as the difference between maximum and minimum marker distance normalized by the respective distance in the static trial. Since the value distribution in four out of 15 parameters (including 5 time– distance parameters) showed significant differences to normal distribution, Wilcoxon signed rank tests were performed between the conditions no shoe–shoe1 and shoe1–shoe2 for all parameters. p-Values are given and indicated by stars when they exceeded the Bonferronicorrected significance level of p < 0.0033.
3. Results The foot kinematics measured for the 18 children are illustrated in Fig. 4. An obvious difference between a typical ‘‘barefoot pattern’’ and a ‘‘shoe pattern’’ was observed. The variation between the subjects (SEM) for both conditions was small compared to the differences between the two measurement conditions ‘shoe1’ and ‘no shoe’. Specifically the following differences were found: The parameter ‘‘tibia foot flexion’’ (Fig. 4(A)) resembled findings of conventional gait analysis [12], but the data overall were shifted towards dorsiflexion by about 58 with a noticeable differences in slope between the two walking conditions. In the shoe, maximum plantarflexion appeared delayed both at loading response and at initial swing. Results were similar in tibio-talar flexion (B) with a significant difference in ROM between the conditions no shoe – shoe1 (22.58 versus 26.68, Table 4). Subtalar rotation (E) was hardly influenced by the footwear showing inversion towards the end of stance phase. Foot torsion (F) and forefoot supination (G) were closely related since they both describe forefoot motion in the frontal plane. In the barefoot measurements (red), a pronating motion is seen at the end of
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
55
Fig. 4. Foot and ankle kinematics: Mean values (heavy lines) across the 18 children given for the conditions no shoe (red) and shoe1 (blue). Upper and lower lines indicate 1 S.E.M. Angles are given in degrees or relative distances in % deviation from the static trial.
stance phase of about 38 when looking along the long foot axis (foot torsion) and similarly when looking along the midfoot axis (forefoot supination). In both parameters, the foot kinematics appeared to change significantly within the shoe (blue) since supination occured at the end of stance phase. Foot progression (H) resembled findings of conven-
tional gait analysis with a larger range of motion without shoes (14.68 versus 11.58). Foot rotation (I), defined as the motion of the long foot axis relative to the knee axis in the transverse plane showed increased motion unexpectedly both with and without shoes. The two distance measures, medial arch length (C) and forefoot width (J) in the barefoot
56
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
Table 4 Mean and S.E.M. of the ROM distribution for the 18 children under conditions no shoe, shoe1, shoe2 Range of motion (8) or variation in distance (%)
No shoe Mean
S.E.M.
Shoe2
Mean
S.E.M.
Mean
Wilcoxon signed rank test S.E.M.
No shoe–shoe1
Shoe1–shoe2
30.0 22.5 9.9 37.1 6.5 9.8 8.4 14.6 20.9 9.7
4.6 3.7 2.5 5.1 1.3 3.0 2.7 4.3 3.9 3.1
29.2 26.6 5.9 25.7 7.4 4.7 6.5 11.5 18.7 4.3
3.7 3.2 1.5 3.5 1.8 1.6 2.1 5.0 4.3 1.4
29.0 25.7 6.0 27.8 7.6 5.2 6.5 12.1 19.4 5.9
4.0 3.3 1.8 4.0 1.7 2.0 1.8 5.1 4.7 1.4
0.446 0.001* <0.001* <0.001* 0.102 <0.001* 0.012 0.002* 0.022 <0.001*
0.248 0.586 0.811 0.014 0.296 0.306 0.528 0.248 0.286 0.002*
1.17 0.91 1.29 132.2 61.7
0.10 0.06 0.14 8.9 1.0
1.24 0.98 1.28 123.5 64.0
0.09 0.06 0.13 7.6 1.1
1.23 0.94 1.31 127.6 63.9
0.11 0.06 0.15 7.56 1.1
0.001* <0.001* 0.679 <0.001* <0.001*
0.862 0.695 0.012 0.777 0.396
(A) Tibia-foot flexion (8) (B) Tibio-talar flexion (8) (C) Medial arch length (%) (D) Hallux flexion (8) (E) Subtalar rotation (8) (F) Foot torsion (8) (G) Forefoot supination (8) (H) Foot progression (8) (I) Foot rotation (8) (J) Forefoot width (%) Stride length (m) Stride time (m) Velocity (m/s) Cadence (steps/s) Toe-off (% gait cycle)
Shoe1
Significant differences between the conditions are indicated with an asterisk when the p-value is below the adjusted level of significance due to multiple parameter testing.
condition showed shortening of the medial arch in late stance by 8% and widening of the forefoot under load in midstance by 8%. They were both influenced by the footwear as these changes reduced to 6% and 2%, respectively. Finally, hallux flexion (D) showed dorsiflexion in late stance when walking barefoot as well as in shoes. The differences in midstance (88 plantar flexed walking barefoot versus 208 dorsiflexed in the shoe) were mainly due to the fact that the toe marker had to be placed onto the shoe. Time–distance parameters where influenced by the footwear. In shoe1 stride length and stride time were increased whereas cadence decreased such that the walking velocity was unchanged. No significant differences were found between the two types of shoes. Overall, 10 out of the 15 parameters displayed in Table 4 showed significant changes due to the footwear when comparing to the barefoot values. Between the two shoe types, only the variation in forefoot width during the gait cycle showed a difference which reached statistical significance. This could lead to the conclusion that the influence of the two shoes onto the foot kinematics was similar. However, the mean values in 12 out of 15 parameters in Table 4 indicate a trend in shoe2 towards the barefoot walking pattern.
For validation, parameters of the HFMM and this model have been compared by calculating correlation coefficients between corresponding parameters including all subjects (Table 5). In addition, mean standard deviations were indicated in degrees. Tibio-talar (B) and hallux flexion (D) show a high correlation (>0.9) to the HFMM corroborating their validity. The weaker correlation coefficient of 0.804 for subtalar rotation (E) is mainly explained by the different definition of the axis of rotation which nonetheless does not compromise the main characteristics of this motion (Fig. 5). Foot torsion (F), supination (G) and rotation (I) do not have a direct counterpart in HFMM and correlation coefficients therefore are small. Tibia-foot flexion (Fig. 4(A)) as well as foot progression (Fig. 4(H)) are not calculated in HFMM, except for possible offset effects these should closely resemble conventional methods to determine ankle dorsiflexion and foot alignment [12].
Table 5 Correlation coefficients between parameters of this model and corresponding parameters of HFMM including all subjects This model
HFMM
CC
s
(B) Tibio-talar flexion (D) Hallux flexion (E) Subtalar rotation (F) Foot torsion (G) Forefoot supination (I) Foot rotation
Tibio-talar flexion Hallux flexion Subtalar inversion Forefoot/ankle supination Forefoot/midfoot supination Forefoot/hindfoot abduction
0.932 0.952 0.804 0.513 0.426 0.774
2.16 4.53 1.81 2.34 1.95 4.17
Mean standard deviations across the gait cycle are indicated in degrees (8).
Fig. 5. Subtalar rotation with HFMM (dashed) and this model (solid) averaged across all 18 children. Mean values (heavy lines) and standard deviations are indicated.
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
4. Discussion The question of optimal footwear choice has a long history and is important especially for the selection of children’s shoes. This choice is not only influenced by health and protection issues but also price and fashion trends. Inorder to increase awareness in this respect, it is necessary to demonstrate evidence of the footwear’s effects on foot shape and to study the direct consequences of the shoe on foot motion and function. In recent years, an increasing number of scientific contributions dealt with foot motion monitored by optical 3D-tracking of skin mounted markers. This method was proven a valuable tool in understanding foot motion [5–11,13–16]. It would therefore be reasonable to also use this technique for the investigation of shod walking in children. Necessary cut-outs in the shoe were kept small to minimize possible changes in their physical properties, but large enough to allow marker movement within the cut-out without any contact with the rim. The mechanical properties were checked subjectively by the ease with which the shoe could manually be twisted about its long axis and flexed at different sections of the shoe. The commercial shoe (shoe1) appeared quite stiff in this respect and the holes hardly had any influence on this stiffness except for the toe marker where it was decided not to use a cut-out as mentioned above. The marker was therefore placed onto the shoe, with the disadvantage that differences in hallux flexion between barefoot and shod walking have to be interpreted with care. Besides this specific aspect, the marker placement used in this work was proven reliable [10]. In preliminary measurements following the methods described above, forefoot motion (parameters F and J in Table 4) appeared diminished in shoe1 and the medial arch length changes (parameter C) were reduced. These results together with suggestions from literature [1] were taken into account for the development of shoe2. This was of the same type as shoe1 but with a slimmer outer sole, a more flexible upper and overall lighter. Since the marker set used in this model is a subset of the one used for the HFMM [10] the accuracy and reliability regarding skin motion would be the same. For the parameter ‘‘medial arch’’ derived in the HFMM, the direct distance between the medial calcaneus marker and the distal MT1 marker, denoted as ‘‘medial arch length’’, has been determined instead (Fig. 4(C)). Correspondingly, instead of the ‘‘MTI-5 angle’’, the distance between MT1 and MT5 defined as ‘‘forefoot width’’ (Fig. 4(J)) has been determined. Hence, both of these parameters would be a valid representation of foot motion. Due to different definitions of neutral position and slightly differing axes and projections, the angular parameters derived from this model in principle differ from those in the HFMM but the slope was characteristically similar. Foot torsion (F), supination (G) and rotation (I) have little correlation to parameters in HFMM. As technical dimensions however, they provide additional information about foot motion.
57
Time–distance parameters (Table 4) as a global measure to monitor the walking process gave further insight into differences between the three walking conditions. Children were asked to walk at their most comfortable speed and no significant differences in the walking velocity were found between the three conditions. The stride length was significantly greater with shoes than without, with no difference between the shoe types. The longer stride length in shoes is in agreement with a previous study [17]. The cadence decreased significantly in shoes compensating for the increased stride length for a comfortable walking velocity which was unchanged. Stance duration (time at toe-off) was increased in comparison with the barefoot condition possibly because the heel strike is dampened and starts somewhat earlier. For the angular parameters, the neutral position was defined by a static capture while standing. Ideally, for this capture the feet should have been placed side by side. However, due to marker occlusion the children were asked to place their heels one foot apart from each other with the feet rotated externally to each other by about 608. Hence, the ankle was slightly plantar-flexed resulting in a shift of the data in tibia-foot flexion and tibio-talar flexion (Fig. 4(A and B)). This may explain why hallux flexion (Fig. 4(D)) in the barefoot trials did not reach neutral position in stance: children may have plantar-flexed their big toe (approximately 78) in the static capture leading to a corresponding offset in the kinematics. Tibia-foot flexion (Fig. 4(A)) which closely corresponds to ankle dorsiflexion in conventional gait analysis together with foot rotation (Fig. 4(I)) described the motion of the foot as a whole relative to the tibia. These parameters were little influenced by the footwear; a finding which is in line with a previous study [17]. Tibio-talar ROM however, which specifically measures hindfoot motion increased significantly when wearing shoe1 (Table 4, B). This can only be understood as a compensation for possibly reduced sagittal plane motion of the mid- and forefoot segments within the shoe since the over-all movement (tibia-foot) was almost identical. Shoe2 did not improve this fact significantly. Barefoot results of subtalar rotation (Fig. 4(E); Table 4, E) confirmed the findings of Carson et al. [9] with a ROM of about 58 and the largest eversion at mid-stance. The kinematics in this case were not influenced by the footwear which was expected. Pronounced ROM of 208 in foot rotation (Table 4, I) however, was not expected as the contribution of subtalar motion in the transverse plane should have been smaller. Hence, transverse mid- and forefoot motion may be important contributing factors. Stebbins et al. [11] and Simon et al. [10] found forefoot ROM of about 108 which does explain the peak in foot rotation (or rather forefoot adduction) at toe-off (Fig. 4(I)). The slope in foot rotation in early stance, however, is probably not explained by subtalar motion alone and might reveal a shortcoming in the definition of the long foot axis since no marker could be used in the region of the second metatarsal. Foot progression (Fig. 4(H)) describing the
58
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
orientation of this long foot axis relative to the gait direction resembled typical findings of conventional gait analysis, but the values appeared shifted internally by about 108. The rather rough and arbitrary definition of the forefoot center located on a weighted mean position (60/40) between first and fifth metatarsal head (Fig. 2, left) obviously led to a medial position. Foot torsion (Fig. 4(F)) and forefoot supination (Fig. 4(G)) were closely related as mentioned above. Nevertheless, these parameters refer to a different axis of rotation (long foot axis versus midfoot axis) and to different markers. Both express the orientation of the metatarsals either in relation to the calcaneus (foot torsion) or to the ankle (forefoot supination) axis. For the barefoot condition, the differences between foot torsion and forefoot supination during loading response were small. The differences in mid- and late stance indicate motion of the calcaneus towards a varus position. Interestingly, the shoe influenced these two parameters differently: along the long foot axis, foot torsion was reduced, possibly due to coupling of calcaneal and forefoot motion. Forefoot supination, however, showed a distinct motion pattern, different to the barefoot condition. Without shoes, the foot pronated to stabilize the forefoot lever arm for push-off [18]. In contrast, this was not observed during in-shoe foot condition (Fig. 4(G)), possibly because forefoot motion was inhibited by the shoe. Consequently, the ROM for foot torsion across the 18 children was significantly reduced within the shoes with little differences between the shoe types (Table 4). The length of the medial arch, defined as the distance between the medial calcaneus marker and the MT1 marker, showed a distinct pattern in the barefoot condition (Fig. 4(C)). Due to the windlass mechanism the medial arch lifted to produce a rigid forefoot lever arm at push-off. The increase in height of the medial arch was visible in a decrease in arch length of about 10% on average in the course of stance phase. In shoe1 this variation in arch length is significantly reduced to about 6% with little differences to shoe2. It appears that the shoe partly inhibited this windlass mechanism. This would go along with the missing forefoot pronation at push-off and the reduced mid and forefoot motion in the sagittal plane as discussed above. The forefoot width, characterising a metatarsal spreading under load in mid-stance showed a variation of almost 10% within the gait cycle for the barefoot condition (Fig. 4(J); Table 4, J). This variation was drastically reduced in the shoes possibly due to a width limitation caused by the shoes. Shoe2 showed a significant improvement in this case with a variation in forefoot width of 6% compared to 4% in shoe1, even though both shoes had the same inner width at the forefoot.
5. Conclusion The present study demonstrated that it is possible to measure the influence of footwear on foot motion offering an
objective basis for functionally testing children shoes. The commercial shoe had a significant influence on the motion patterns particularly at the forefoot. The experimental shoe developed in the course of the study with a slimmer and more flexible design showed a tendency to reduce this influence and was closer to barefoot motion pattern. However, it was not a ‘‘neutral’’ shoe and did influence motion patterns, such as the medial arch motion. Nevertheless, the encouraging findings with this shoe together with previous recommendations by Staheli and Maier [1,19] would support the principle: ‘‘The shoe should in no other way influence the normal foot than to protect it against lesion and coldness.’’
Conflict of interest The author and all of the co-authors hereby certify that they have no financial interest in relation with this publication.
Acknowledgment The financial support of the company Elefanten GmbH, Kleve, Germany, is gratefully acknowledged.
References [1] Staheli LT. Shoes for children: a review. Pediatrics 1991;88(2):371–5. [2] Rao UB, Joseph B. The influence of footwear on the prevalence of flat foot. A survey of 2300 children. J Bone Joint Surg Br 1992;74(4):525– 7. [3] Jerosch J, Mamsch H. Deformities and misalignment of feet in children—a field study of 345 students. Z Orthop Ihre Grenzgeb 1998;136(3):215–20. [4] Wearing SC, Smeathers JE, Yates B, Sullivan PM, Urry SR, Dubois P. Errors in measuring sagittal arch kinematics of the human foot with digital fluoroscopy. Gait Posture 2005;21(3):326–32. [5] Leardini A, Benedetti MG, Catani F, Simoncini L, Giannini S. An anatomically based protocol for the description of foot segment kinematics during gait. Clin Biomech 1999;14(8):528–36. [6] MacWilliams BA, Cowley M, Nicholson DE. Foot kinematics and kinetics during adolescent gait. Gait Posture 2003;17(3):214–24. [7] Myers KA, Wang M, Marks RM, Harris GF. Validation of a multisegment foot and ankle kinematic model for pediatric gait. IEEE Trans Neural Syst Rehabil Eng 2004;12(1):122–30. [8] Arampatzis A, Bruggemann GP, Klapsing GM. A three-dimensional shank-foot model to determine the foot motion during landings. Med Sci Sports Exerc 2002;34(1):130–8. [9] Carson MC, Harrington ME, Thompson N, O’Connor JJ, Theologis TN. Kinematic analysis of a multi-segment foot model for research and clinical applications: a repeatability analysis. J Biomech 2001;34(10):1299–307. [10] Simon J, Doederlein L, McIntosh AS, Metaxiotis D, Bock HG, Wolf SI. The Heidelberg foot measurement method: development, description and assessment. Gait Posture 2006;23(4):411–24. [11] Stebbins J, Harrington M, Thompson N, Zavatsky A, Theologis T. Repeatability of a model for measuring multi-segment foot kinematics in children. Gait Posture 2006;23(4):401–10. [12] Kadaba MP, Ramakrishnan HK, Wootten ME, Gainey J, Gorton G, Cochran GVB. Repeatability of kinematic, kinetic, and electro-
S. Wolf et al. / Gait & Posture 27 (2008) 51–59 myographic data in normal adult gait. J Orthop Res 1989;7(6):849– 60. [13] Cornwall MW, McPoil TG. Motion of the calcaneus, navicular, and first metatarsal during the stance phase of walking. J Am Podiatr Med Assoc 2002;92(2):67–76. [14] Kidder SM, Abuzzahab Jr FS, Harris GF, Johnson JE. A system for the analysis of foot and ankle kinematics during gait. IEEE Trans Rehabil Eng 1996;4(1):25–32. [15] Scott SH, Winter DA. Biomechanical model of the human foot: kinematics and kinetics during the stance phase of walking. J Biomech 1993;26(9):1091–104.
59
[16] Woodburn J, Nelson KM, Siegel KL, Kepple TM, Gerber LH. Multisegment foot motion during gait: proof of concept in rheumatoid arthritis. J Rheumatol 2004;31(10):1918–27. [17] Oeffinger D, Brauch B, Cranfill S, Hisle C, Wynn C, Hicks R, et al. Comparison of gait with and without shoes in children. Gait Posture 1999;9(2):95–100. [18] Perry JP. Gait analysis: normal and pathological function. Thorofare, NJ: Slack; 1992. [19] Der Kinderschuh Maier E. In: Baumgartner R, Stinus H, editors. Die orhopa¨dietechnische Versorgung des Fußes. Stuttgart: Thieme Verlag; 1995. p. 163–9.
Foot motion in children shoes—A comparison of barefoot walking with shod walking in conventional and flexible shoes Sebastian Wolf a,*, Jan Simon b, Dimitrios Patikas a, Waltraud Schuster a, Petra Armbrust a, Leonhard Do¨derlein a b
a Department of Orthopedic Surgery, University of Heidelberg, Schlierbacher Landstrasse 200a, 69118 Heidelberg, Germany Interdisciplinary Center for Scientific Computation, Heidelberg (IWR), Im Neuenheimer Feld 368, 69120 Heidelberg, Germany
Received 13 July 2006; received in revised form 11 January 2007; accepted 13 January 2007
Abstract The increased prevalence for flatfoot and hallux valgus in modern societies may be the consequence of inaequate footwear in childhood. Based on the assumption that barefoot walking represents the best condition for the development of a healthy foot the objective of this study was to monitor the influence of commercial footwear on children’s foot motion during walking. Furthermore, an attempt was made to reduce this influence by changing the physical properties of standard footwear. Children’s barefoot motion pattern was monitored by a marker-based optical 3D-tracking method using a multi-segment foot model. In the study’s first stage, barefoot walking was compared to walking with a commercial product. In the second stage it was compared to both, the pattern with the commercial product and with the shoe modified on the basis of the findings of the first stage of the study. Eighteen children (8.2 0.7 years old) with no foot deformity and with the same shoe size were recruited for this study. It was found that tibio-talar ROM increased in the commercial shoe (26.68) compared to the barefoot condition (22.58, p = 0.001) whereas the medial arch changes for push-off were diminished since the variation in arch length was reduced from 9.9% (barefoot) to 5.9% (shoe, p < 0.001). Further, ROM in foot torsion along the long foot axis was reduced from 9.88 (bare) to 4.78 (shoe, p < 0.001). These parameters could be improved with more flexible footwear. The present study shows that slimmer and more flexible children’s shoes do not change foot motion as much as conventional shoes and therefore should be recommended not only for children in this age but for healthy children in general. # 2007 Elsevier B.V. All rights reserved. Keywords: Footwear; Children; Motion analysis
1. Introduction Shoes are primarily used to protect the foot from injuries due to rough or uneven ground surfaces and from excessive impact due to a hard ground. Furthermore, shoes protect the feet from a cold and wet environment. This holds true, in principle, for children’s shoes as well. However, it can be claimed, that optimum foot development can only occur in barefoot conditions. Stiff and tight footwear may lead to deformity and stiffness [1]. The design of children’s shoes should be based on the barefoot model taking into account shock absorption and load distribution [1]. There are several * Corresponding author. Tel.: +49 6221 966724; fax: +49 6221 966725. E-mail address: [email protected] (S. Wolf). 0966-6362/$ – see front matter # 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2007.01.005
studies indicating that inadequate footwear or even footwear in general may affect the physiological development of the foot. Rao and Joseph [2] showed that in 2300 children between the age of 4 and 13 years the incidence of a flat foot among those who used footwear (8.6%) was significantly higher compared to those who did not (2.8%). This effect was found in all age groups. Moreover, children wearing closed shoes showed a higher incidence for flat feet than those wearing sandals or slippers. Jerosch and Mamsch [3] documented that in children between 10 and 13 years of age only 36.5% had normal feet. The rest had mild to significant deformities, usually valgus (39.4%). Flat foot deformity was diagnosed in 19.1% and a hallux valgus in 17.1% of the children. Little is known about the direct effects of foot wear on the kinematic behaviour of a child’s foot. Fluoroscopic
52
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
measurements are possible [4], but ethically inappropriate in children. Optical 3D tracking of markers placed on the foot is a valuable tool for the analysis of foot kinematics [5–10] which is feasible also for children [11] but, to our knowledge, has never been applied to shod walking in children. The aim of this study was to apply a marker-based optical method to bare foot walking as well as shod walking by perforating the shoe at locations at which markers were to be applied directly on the foot. The marker protocol used in the Heidelberg Foot Measurement Method (HFMM) [10] has therefore been reduced and the model simplified to monitor the child’s foot movement ‘‘barefoot’’ within the shoe. The aim of the first stage of this prospective study was to assess the influence of a typical and commercially available children’s shoe (shoe1) on foot motion. The second stage aimed at developing a modified shoe (shoe2) based on the findings of stage one as well as the recommendations by Staheli [1]: his study suggested that a thinner and more flexible sole and upper would allow better afferent stimulus as well as better mobility of foot segments. In order to assess the new shoe design, the foot motion patterns in shoe1 and shoe2 were compared to the barefoot pattern.
2. Participants and methods 2.1. Data collection In the recruitment process, more than 100 children between 6 and 10 years of age were initially assessed. Eighteen children (age 8.2 0.7 years, 8 girls, 10 boys) with no foot deformities, previous significant injuries or operations to their feet were elected for the study. All children were fitted with shoes of French size 33. Foot size effects in this small distribution of foot lengths (20.7 0.4 cm) could be eliminated. All children and parents/carers consented according to the local ethics committee regulations. Ten reflective markers of 14 mm diameter were placed on the left shank and foot as shown in Figs. 1 and 2 for the foot kinematic measurements. This marker setup consists of a subset of markers used for the HFMM [10] excluding three markers at the mid-foot. In the barefoot trials, the complete marker set of the HFMM has been used for validation reasons.
Table 1 Physical properties of the shoes Property
Shoe1
Shoe2
Weight (g) Total length (cm) Inner length (cm) Forefoot outer width (cm) Forefoot inner width (cm) Sole thickness heel (mm) Sole thickness forefoot (mm) Profile thickness
245 23.5 21.9 8.9 7.5 23 14 3.0
158 23.2 21.9 8.6 7.5 7 7 1.5
Shoe1: commercial shoe, shoe2: experimental shoe.
Shoe1, a commercial shoe (Elefanten, Germany; Fig. 1, right) with Balmoral closure style was prepared with four additional cut-outs for placing markers at first and fifth metatarsal and the medial and lateral sides of the calcaneus. The markers were placed at the same spot onto the skin as in the barefoot measurements except for the hallux marker which was placed onto the shoe. Without further manipulation, the shoe left enough room to place the malleoli markers onto the skin. Heel and toe markers on both feet were used for event detection. The experimental shoe2, prepared in a similar manner, had a slimmer outer sole, a more flexible upper and was overall lighter (Fig. 3). The insole was identical. Some physical properties are listed in Table 1. In the beginning of each test session, a static trial with the barefoot marker placement was captured in standing position as reference. The children were then asked to walk on a 7 m walkway at self-selected speed monitored by a motion capture system (6 cameras VICON 612 @ 120 Hz). The measurement was repeated with shoe1 and shoe2. The positions of the markers were marked with a felt tip pen on the skin to preserve identical marker positioning after wearing or changing shoes. If the prepared cut-outs in the shoe did not allow an exact placement of a marker, the cutouts were slightly enlarged without compromising the stability of the shoe. Care was taken to avoid any contact between the markers and the rim of the cut-outs. 2.2. Calculation of foot motion parameters In order to measure foot motion, four additional virtual positions were calculated to represent anatomical reference points in the shank and foot: (1) a center point in the calcaneus as midpoint between the two calcaneus markers,
Fig. 1. Marker placement barefoot (left) and in the shoe1 (right). Four cut-outs were used for the medial and lateral calcaneal and metatarsal markers.
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
53
Fig. 2. Markers used for the measurement—LEP, MEP: lateral and medial femoral epicondyle; LML, MML: lateral and medial malleolus; LCL, MCL: lateral, medial calcaneus; MT1, MT5: first (medial) and fifth (lateral) metatarsal head; HLX: distal phalanx of hallux. CCL: heel marker for event detection. Additional virtual points and lines for the model calculation of joint centers, axes and segment lines are indicated.
(2) a forefoot center point defined as a weighted mean position between first (60%) and fifth (40%) metatarsal head marker, (3) the ankle center as the midpoint between the malleoli and (4) the knee center as the midpoint between medial and lateral femoral epicondyles (Fig. 2). Connecting lines between these four reference points in the sagittal plane represent the long axis of the tibia and three axes within the foot: hindfoot axis, midfoot axis and long foot axis (Fig. 2,
right). The connecting line between the physical markers on the malleoli defined the ankle joint axis; metatarsal and calcaneus markers defined a forefoot axis and a calcaneus line, respectively (Table 2). Rotation angles within the foot were defined between pairs of line-like 2D-segments represented by these connecting lines. For simplicity, the same lines were also used to define rotation axes articulating the 2D-segments. The hindfoot axis for example represents the orientation of the calcaneus relative to the long axis of the tibia describing ankle flexion (tibio-talar flexion, Table 3) about the ankle axis i.e. the tibio-talar joint. Further, the hindfoot axis served to define a rotation axis for subtalar joint motion between hindfoot and tibia (Table 3) roughly coinciding with the biomechanical subtalar joint axis. Technically, rotation between pairs of 2D-segments was determined as projection angle by projecting both segments into a plane perpendicular to the rotation axis as previously Table 2 Lines and axes defined by locations of real and virtual marker positions
Fig. 3. Sole of experimental shoe2 (top) and lateral view (bottom).
No.
Line/axis
Marker
1 2 3 4 5 6 7 8 9 10 11
Knee axis Ankle axis Tibia axis Long foot axis Midfoot axis Hind foot axis Calcaneus line Medial arch line Forefoot axis Hallux line Gait direction
MEP–LEP MML–LML Knee center–ankle center Calcaneus center–forefoot center Ankle center–forefoot center Ankle center–calcaneus center MCL–LCL MCL–MT1 MT1–MT5 MT1–HLX CCL (between heel strikes)
54
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
Table 3 Definition of angles between pairs of segments No.
Angle
Axis of rotation
Segment 1
Segment 2
1 2 3 4 5 6 7 8
Tibia-foot flexion Tibio-talar flexion Subtalar rotation Hallux flexion Foot rotation Foot torsion Forefoot supination Foot progression
Ankle axis Ankle axis Hind foot axis Forefoot axis Tibia axis Long foot axis Midfoot axis Tibia axis
Tibia axis Tibia axis Calcaneus line Midfoot axis Long foot axis Forefoot axis Forefoot axis Long foot axis
Long foot axis Hind foot axis Ankle axis Hallux line Knee axis Hind foot axis Ankle axis Gait direction
reported [10]. This was calculated as the scalar product of two vector products e.g.
Furthermore, the time–distance parameters were determined for each subject and condition.
nankle-axis ~ nhindfoot-axis Þ; cosðasubtalar-rotation Þ ¼ hð~ ð~ ncalcaneus-line ~ nhindfoot-axis Þi:
2.3. Statistics (1)
The vectors in Eq. (1) are normalized direction vectors. In the static trial, the anatomic neutral position was defined for all angles. Tibia-foot flexion (no. 1 in Table 3) differed and was calculated directly in the plane defined by the lines of the tibia and the long foot axis which did not perfectly coincide with the normal plane of the ankle axis. The numerical results however were very similar. For further analysis, the unfiltered data of the foot kinematics were time-normalized to the gait cycle and averaged across 7–10 strides for each child and measurement condition. Seven angles describing the orientation between different foot segments were calculated according to Table 3: (1) tibia-foot flexion, in conventional gait analysis referred to as ‘‘ankle flexion’’ or ‘‘dorsiflexion’’ which measures sagittal plane motion of the foot (regarded as one rigid segment) relative to the tibia, (2) tibio-talar flexion, which measures the same kind of movement but refers to markers on tibia and hindfoot only such that mid and forefoot motion is excluded, (3) subtalar rotation, measuring hindfoot motion relative to the tibia about an axis which is close to the functional axis of the subtalar joint, (4) hallux flexion, which addresses sagittal plane motion of the first toe, (5) foot rotation measuring motion of the complete foot relative to the tibia similar to tibia-foot flexion but in the transverse plane, (6) foot torsion, a rather technical parameter measuring transverse plane forefoot motion relative to the hindfoot, and (7) forefoot supination, measuring forefoot motion relative to the ankle. Further, foot progression (8), the orientation of the long foot axis with respect to the gait direction is determined. For validation, correlations with corresponding parameters of the HFMM were calculated. The distance between the first metatarsal marker (MT1) and the medial calcaneus marker (MCL) representing the length of the medial arch was also measured. The relative change to this distance during the static trial was calculated in %. Similarly, the change in forefoot width was estimated by calculating the distance between the markers of first and fifth metatarsal (MT1 and MT5), respectively (Table 3).
For quantification of differences between the measurement conditions, the individual range of motion (ROM) for each angular parameter was determined. The change in medial arch length and forefoot width within the gait cycle (given in %) was calculated as the difference between maximum and minimum marker distance normalized by the respective distance in the static trial. Since the value distribution in four out of 15 parameters (including 5 time– distance parameters) showed significant differences to normal distribution, Wilcoxon signed rank tests were performed between the conditions no shoe–shoe1 and shoe1–shoe2 for all parameters. p-Values are given and indicated by stars when they exceeded the Bonferronicorrected significance level of p < 0.0033.
3. Results The foot kinematics measured for the 18 children are illustrated in Fig. 4. An obvious difference between a typical ‘‘barefoot pattern’’ and a ‘‘shoe pattern’’ was observed. The variation between the subjects (SEM) for both conditions was small compared to the differences between the two measurement conditions ‘shoe1’ and ‘no shoe’. Specifically the following differences were found: The parameter ‘‘tibia foot flexion’’ (Fig. 4(A)) resembled findings of conventional gait analysis [12], but the data overall were shifted towards dorsiflexion by about 58 with a noticeable differences in slope between the two walking conditions. In the shoe, maximum plantarflexion appeared delayed both at loading response and at initial swing. Results were similar in tibio-talar flexion (B) with a significant difference in ROM between the conditions no shoe – shoe1 (22.58 versus 26.68, Table 4). Subtalar rotation (E) was hardly influenced by the footwear showing inversion towards the end of stance phase. Foot torsion (F) and forefoot supination (G) were closely related since they both describe forefoot motion in the frontal plane. In the barefoot measurements (red), a pronating motion is seen at the end of
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
55
Fig. 4. Foot and ankle kinematics: Mean values (heavy lines) across the 18 children given for the conditions no shoe (red) and shoe1 (blue). Upper and lower lines indicate 1 S.E.M. Angles are given in degrees or relative distances in % deviation from the static trial.
stance phase of about 38 when looking along the long foot axis (foot torsion) and similarly when looking along the midfoot axis (forefoot supination). In both parameters, the foot kinematics appeared to change significantly within the shoe (blue) since supination occured at the end of stance phase. Foot progression (H) resembled findings of conven-
tional gait analysis with a larger range of motion without shoes (14.68 versus 11.58). Foot rotation (I), defined as the motion of the long foot axis relative to the knee axis in the transverse plane showed increased motion unexpectedly both with and without shoes. The two distance measures, medial arch length (C) and forefoot width (J) in the barefoot
56
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
Table 4 Mean and S.E.M. of the ROM distribution for the 18 children under conditions no shoe, shoe1, shoe2 Range of motion (8) or variation in distance (%)
No shoe Mean
S.E.M.
Shoe2
Mean
S.E.M.
Mean
Wilcoxon signed rank test S.E.M.
No shoe–shoe1
Shoe1–shoe2
30.0 22.5 9.9 37.1 6.5 9.8 8.4 14.6 20.9 9.7
4.6 3.7 2.5 5.1 1.3 3.0 2.7 4.3 3.9 3.1
29.2 26.6 5.9 25.7 7.4 4.7 6.5 11.5 18.7 4.3
3.7 3.2 1.5 3.5 1.8 1.6 2.1 5.0 4.3 1.4
29.0 25.7 6.0 27.8 7.6 5.2 6.5 12.1 19.4 5.9
4.0 3.3 1.8 4.0 1.7 2.0 1.8 5.1 4.7 1.4
0.446 0.001* <0.001* <0.001* 0.102 <0.001* 0.012 0.002* 0.022 <0.001*
0.248 0.586 0.811 0.014 0.296 0.306 0.528 0.248 0.286 0.002*
1.17 0.91 1.29 132.2 61.7
0.10 0.06 0.14 8.9 1.0
1.24 0.98 1.28 123.5 64.0
0.09 0.06 0.13 7.6 1.1
1.23 0.94 1.31 127.6 63.9
0.11 0.06 0.15 7.56 1.1
0.001* <0.001* 0.679 <0.001* <0.001*
0.862 0.695 0.012 0.777 0.396
(A) Tibia-foot flexion (8) (B) Tibio-talar flexion (8) (C) Medial arch length (%) (D) Hallux flexion (8) (E) Subtalar rotation (8) (F) Foot torsion (8) (G) Forefoot supination (8) (H) Foot progression (8) (I) Foot rotation (8) (J) Forefoot width (%) Stride length (m) Stride time (m) Velocity (m/s) Cadence (steps/s) Toe-off (% gait cycle)
Shoe1
Significant differences between the conditions are indicated with an asterisk when the p-value is below the adjusted level of significance due to multiple parameter testing.
condition showed shortening of the medial arch in late stance by 8% and widening of the forefoot under load in midstance by 8%. They were both influenced by the footwear as these changes reduced to 6% and 2%, respectively. Finally, hallux flexion (D) showed dorsiflexion in late stance when walking barefoot as well as in shoes. The differences in midstance (88 plantar flexed walking barefoot versus 208 dorsiflexed in the shoe) were mainly due to the fact that the toe marker had to be placed onto the shoe. Time–distance parameters where influenced by the footwear. In shoe1 stride length and stride time were increased whereas cadence decreased such that the walking velocity was unchanged. No significant differences were found between the two types of shoes. Overall, 10 out of the 15 parameters displayed in Table 4 showed significant changes due to the footwear when comparing to the barefoot values. Between the two shoe types, only the variation in forefoot width during the gait cycle showed a difference which reached statistical significance. This could lead to the conclusion that the influence of the two shoes onto the foot kinematics was similar. However, the mean values in 12 out of 15 parameters in Table 4 indicate a trend in shoe2 towards the barefoot walking pattern.
For validation, parameters of the HFMM and this model have been compared by calculating correlation coefficients between corresponding parameters including all subjects (Table 5). In addition, mean standard deviations were indicated in degrees. Tibio-talar (B) and hallux flexion (D) show a high correlation (>0.9) to the HFMM corroborating their validity. The weaker correlation coefficient of 0.804 for subtalar rotation (E) is mainly explained by the different definition of the axis of rotation which nonetheless does not compromise the main characteristics of this motion (Fig. 5). Foot torsion (F), supination (G) and rotation (I) do not have a direct counterpart in HFMM and correlation coefficients therefore are small. Tibia-foot flexion (Fig. 4(A)) as well as foot progression (Fig. 4(H)) are not calculated in HFMM, except for possible offset effects these should closely resemble conventional methods to determine ankle dorsiflexion and foot alignment [12].
Table 5 Correlation coefficients between parameters of this model and corresponding parameters of HFMM including all subjects This model
HFMM
CC
s
(B) Tibio-talar flexion (D) Hallux flexion (E) Subtalar rotation (F) Foot torsion (G) Forefoot supination (I) Foot rotation
Tibio-talar flexion Hallux flexion Subtalar inversion Forefoot/ankle supination Forefoot/midfoot supination Forefoot/hindfoot abduction
0.932 0.952 0.804 0.513 0.426 0.774
2.16 4.53 1.81 2.34 1.95 4.17
Mean standard deviations across the gait cycle are indicated in degrees (8).
Fig. 5. Subtalar rotation with HFMM (dashed) and this model (solid) averaged across all 18 children. Mean values (heavy lines) and standard deviations are indicated.
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
4. Discussion The question of optimal footwear choice has a long history and is important especially for the selection of children’s shoes. This choice is not only influenced by health and protection issues but also price and fashion trends. Inorder to increase awareness in this respect, it is necessary to demonstrate evidence of the footwear’s effects on foot shape and to study the direct consequences of the shoe on foot motion and function. In recent years, an increasing number of scientific contributions dealt with foot motion monitored by optical 3D-tracking of skin mounted markers. This method was proven a valuable tool in understanding foot motion [5–11,13–16]. It would therefore be reasonable to also use this technique for the investigation of shod walking in children. Necessary cut-outs in the shoe were kept small to minimize possible changes in their physical properties, but large enough to allow marker movement within the cut-out without any contact with the rim. The mechanical properties were checked subjectively by the ease with which the shoe could manually be twisted about its long axis and flexed at different sections of the shoe. The commercial shoe (shoe1) appeared quite stiff in this respect and the holes hardly had any influence on this stiffness except for the toe marker where it was decided not to use a cut-out as mentioned above. The marker was therefore placed onto the shoe, with the disadvantage that differences in hallux flexion between barefoot and shod walking have to be interpreted with care. Besides this specific aspect, the marker placement used in this work was proven reliable [10]. In preliminary measurements following the methods described above, forefoot motion (parameters F and J in Table 4) appeared diminished in shoe1 and the medial arch length changes (parameter C) were reduced. These results together with suggestions from literature [1] were taken into account for the development of shoe2. This was of the same type as shoe1 but with a slimmer outer sole, a more flexible upper and overall lighter. Since the marker set used in this model is a subset of the one used for the HFMM [10] the accuracy and reliability regarding skin motion would be the same. For the parameter ‘‘medial arch’’ derived in the HFMM, the direct distance between the medial calcaneus marker and the distal MT1 marker, denoted as ‘‘medial arch length’’, has been determined instead (Fig. 4(C)). Correspondingly, instead of the ‘‘MTI-5 angle’’, the distance between MT1 and MT5 defined as ‘‘forefoot width’’ (Fig. 4(J)) has been determined. Hence, both of these parameters would be a valid representation of foot motion. Due to different definitions of neutral position and slightly differing axes and projections, the angular parameters derived from this model in principle differ from those in the HFMM but the slope was characteristically similar. Foot torsion (F), supination (G) and rotation (I) have little correlation to parameters in HFMM. As technical dimensions however, they provide additional information about foot motion.
57
Time–distance parameters (Table 4) as a global measure to monitor the walking process gave further insight into differences between the three walking conditions. Children were asked to walk at their most comfortable speed and no significant differences in the walking velocity were found between the three conditions. The stride length was significantly greater with shoes than without, with no difference between the shoe types. The longer stride length in shoes is in agreement with a previous study [17]. The cadence decreased significantly in shoes compensating for the increased stride length for a comfortable walking velocity which was unchanged. Stance duration (time at toe-off) was increased in comparison with the barefoot condition possibly because the heel strike is dampened and starts somewhat earlier. For the angular parameters, the neutral position was defined by a static capture while standing. Ideally, for this capture the feet should have been placed side by side. However, due to marker occlusion the children were asked to place their heels one foot apart from each other with the feet rotated externally to each other by about 608. Hence, the ankle was slightly plantar-flexed resulting in a shift of the data in tibia-foot flexion and tibio-talar flexion (Fig. 4(A and B)). This may explain why hallux flexion (Fig. 4(D)) in the barefoot trials did not reach neutral position in stance: children may have plantar-flexed their big toe (approximately 78) in the static capture leading to a corresponding offset in the kinematics. Tibia-foot flexion (Fig. 4(A)) which closely corresponds to ankle dorsiflexion in conventional gait analysis together with foot rotation (Fig. 4(I)) described the motion of the foot as a whole relative to the tibia. These parameters were little influenced by the footwear; a finding which is in line with a previous study [17]. Tibio-talar ROM however, which specifically measures hindfoot motion increased significantly when wearing shoe1 (Table 4, B). This can only be understood as a compensation for possibly reduced sagittal plane motion of the mid- and forefoot segments within the shoe since the over-all movement (tibia-foot) was almost identical. Shoe2 did not improve this fact significantly. Barefoot results of subtalar rotation (Fig. 4(E); Table 4, E) confirmed the findings of Carson et al. [9] with a ROM of about 58 and the largest eversion at mid-stance. The kinematics in this case were not influenced by the footwear which was expected. Pronounced ROM of 208 in foot rotation (Table 4, I) however, was not expected as the contribution of subtalar motion in the transverse plane should have been smaller. Hence, transverse mid- and forefoot motion may be important contributing factors. Stebbins et al. [11] and Simon et al. [10] found forefoot ROM of about 108 which does explain the peak in foot rotation (or rather forefoot adduction) at toe-off (Fig. 4(I)). The slope in foot rotation in early stance, however, is probably not explained by subtalar motion alone and might reveal a shortcoming in the definition of the long foot axis since no marker could be used in the region of the second metatarsal. Foot progression (Fig. 4(H)) describing the
58
S. Wolf et al. / Gait & Posture 27 (2008) 51–59
orientation of this long foot axis relative to the gait direction resembled typical findings of conventional gait analysis, but the values appeared shifted internally by about 108. The rather rough and arbitrary definition of the forefoot center located on a weighted mean position (60/40) between first and fifth metatarsal head (Fig. 2, left) obviously led to a medial position. Foot torsion (Fig. 4(F)) and forefoot supination (Fig. 4(G)) were closely related as mentioned above. Nevertheless, these parameters refer to a different axis of rotation (long foot axis versus midfoot axis) and to different markers. Both express the orientation of the metatarsals either in relation to the calcaneus (foot torsion) or to the ankle (forefoot supination) axis. For the barefoot condition, the differences between foot torsion and forefoot supination during loading response were small. The differences in mid- and late stance indicate motion of the calcaneus towards a varus position. Interestingly, the shoe influenced these two parameters differently: along the long foot axis, foot torsion was reduced, possibly due to coupling of calcaneal and forefoot motion. Forefoot supination, however, showed a distinct motion pattern, different to the barefoot condition. Without shoes, the foot pronated to stabilize the forefoot lever arm for push-off [18]. In contrast, this was not observed during in-shoe foot condition (Fig. 4(G)), possibly because forefoot motion was inhibited by the shoe. Consequently, the ROM for foot torsion across the 18 children was significantly reduced within the shoes with little differences between the shoe types (Table 4). The length of the medial arch, defined as the distance between the medial calcaneus marker and the MT1 marker, showed a distinct pattern in the barefoot condition (Fig. 4(C)). Due to the windlass mechanism the medial arch lifted to produce a rigid forefoot lever arm at push-off. The increase in height of the medial arch was visible in a decrease in arch length of about 10% on average in the course of stance phase. In shoe1 this variation in arch length is significantly reduced to about 6% with little differences to shoe2. It appears that the shoe partly inhibited this windlass mechanism. This would go along with the missing forefoot pronation at push-off and the reduced mid and forefoot motion in the sagittal plane as discussed above. The forefoot width, characterising a metatarsal spreading under load in mid-stance showed a variation of almost 10% within the gait cycle for the barefoot condition (Fig. 4(J); Table 4, J). This variation was drastically reduced in the shoes possibly due to a width limitation caused by the shoes. Shoe2 showed a significant improvement in this case with a variation in forefoot width of 6% compared to 4% in shoe1, even though both shoes had the same inner width at the forefoot.
5. Conclusion The present study demonstrated that it is possible to measure the influence of footwear on foot motion offering an
objective basis for functionally testing children shoes. The commercial shoe had a significant influence on the motion patterns particularly at the forefoot. The experimental shoe developed in the course of the study with a slimmer and more flexible design showed a tendency to reduce this influence and was closer to barefoot motion pattern. However, it was not a ‘‘neutral’’ shoe and did influence motion patterns, such as the medial arch motion. Nevertheless, the encouraging findings with this shoe together with previous recommendations by Staheli and Maier [1,19] would support the principle: ‘‘The shoe should in no other way influence the normal foot than to protect it against lesion and coldness.’’
Conflict of interest The author and all of the co-authors hereby certify that they have no financial interest in relation with this publication.
Acknowledgment The financial support of the company Elefanten GmbH, Kleve, Germany, is gratefully acknowledged.
References [1] Staheli LT. Shoes for children: a review. Pediatrics 1991;88(2):371–5. [2] Rao UB, Joseph B. The influence of footwear on the prevalence of flat foot. A survey of 2300 children. J Bone Joint Surg Br 1992;74(4):525– 7. [3] Jerosch J, Mamsch H. Deformities and misalignment of feet in children—a field study of 345 students. Z Orthop Ihre Grenzgeb 1998;136(3):215–20. [4] Wearing SC, Smeathers JE, Yates B, Sullivan PM, Urry SR, Dubois P. Errors in measuring sagittal arch kinematics of the human foot with digital fluoroscopy. Gait Posture 2005;21(3):326–32. [5] Leardini A, Benedetti MG, Catani F, Simoncini L, Giannini S. An anatomically based protocol for the description of foot segment kinematics during gait. Clin Biomech 1999;14(8):528–36. [6] MacWilliams BA, Cowley M, Nicholson DE. Foot kinematics and kinetics during adolescent gait. Gait Posture 2003;17(3):214–24. [7] Myers KA, Wang M, Marks RM, Harris GF. Validation of a multisegment foot and ankle kinematic model for pediatric gait. IEEE Trans Neural Syst Rehabil Eng 2004;12(1):122–30. [8] Arampatzis A, Bruggemann GP, Klapsing GM. A three-dimensional shank-foot model to determine the foot motion during landings. Med Sci Sports Exerc 2002;34(1):130–8. [9] Carson MC, Harrington ME, Thompson N, O’Connor JJ, Theologis TN. Kinematic analysis of a multi-segment foot model for research and clinical applications: a repeatability analysis. J Biomech 2001;34(10):1299–307. [10] Simon J, Doederlein L, McIntosh AS, Metaxiotis D, Bock HG, Wolf SI. The Heidelberg foot measurement method: development, description and assessment. Gait Posture 2006;23(4):411–24. [11] Stebbins J, Harrington M, Thompson N, Zavatsky A, Theologis T. Repeatability of a model for measuring multi-segment foot kinematics in children. Gait Posture 2006;23(4):401–10. [12] Kadaba MP, Ramakrishnan HK, Wootten ME, Gainey J, Gorton G, Cochran GVB. Repeatability of kinematic, kinetic, and electro-
S. Wolf et al. / Gait & Posture 27 (2008) 51–59 myographic data in normal adult gait. J Orthop Res 1989;7(6):849– 60. [13] Cornwall MW, McPoil TG. Motion of the calcaneus, navicular, and first metatarsal during the stance phase of walking. J Am Podiatr Med Assoc 2002;92(2):67–76. [14] Kidder SM, Abuzzahab Jr FS, Harris GF, Johnson JE. A system for the analysis of foot and ankle kinematics during gait. IEEE Trans Rehabil Eng 1996;4(1):25–32. [15] Scott SH, Winter DA. Biomechanical model of the human foot: kinematics and kinetics during the stance phase of walking. J Biomech 1993;26(9):1091–104.
59
[16] Woodburn J, Nelson KM, Siegel KL, Kepple TM, Gerber LH. Multisegment foot motion during gait: proof of concept in rheumatoid arthritis. J Rheumatol 2004;31(10):1918–27. [17] Oeffinger D, Brauch B, Cranfill S, Hisle C, Wynn C, Hicks R, et al. Comparison of gait with and without shoes in children. Gait Posture 1999;9(2):95–100. [18] Perry JP. Gait analysis: normal and pathological function. Thorofare, NJ: Slack; 1992. [19] Der Kinderschuh Maier E. In: Baumgartner R, Stinus H, editors. Die orhopa¨dietechnische Versorgung des Fußes. Stuttgart: Thieme Verlag; 1995. p. 163–9.