Custom-Molded Foot-Orthosis Intervention and Multisegment Medial Foot Kinematics During Walking

Participants

We recruited apparently healthy people from Georgia State University and the surrounding community. After initially screening them for current musculoskeletal injuries, we further screened potential participants for eligibility through arch height and foot mobility assessment using the arch ratio in 90% weight bearing and the relative arch deformity (RAD) ratio, respectively.13 We enrolled 16 participants (7 men, 9 women) with low-mobile foot posture (Table 1). We defined low arch structure (arch ratio ≤ 0.287) as an arch ratio equal to or greater than 1 SD below the mean arch ratio assessed in 51 random volunteers (102 feet). We defined a mobile foot (RAD ratio > 0.828 10⁴/N) as a RAD ratio greater than the mean ratio of the mean assessed in the same 102 feet (Table 1). For the arch ratio, a smaller ratio is associated with a lower arch, and for the RAD ratio, a larger ratio is associated with a more mobile foot. All participants provided written informed consent, and the Institutional Review Board of Georgia State University approved the study.

Table 1. Demographic Data of Participants (Mean ± SD)

View Fullscreen

Three-Dimensional Motion Analysis

Eight optical video cameras (model TM-6703; PULNiX America, Inc, Sunnyvale, CA) sampling at 120 Hz were used to capture 3-dimensional coordinate data from clusters of 3 or 4 retroreflective markers (8-mm diameter) located on the leg and foot segments of interest. We placed the markers either directly on the skin or mounted on wands constructed from 1.8-mm wire that we fixed to the skin with liquid adhesive and double-sided tape. An AMTI force platform (Advanced Mechanical Technology, Newton, MA) sampling at 960 Hz and an AMTI amplifier (1050-Hz, second-order, critically damped filter with a gain of 1000) were used to determine initial contact and toe-off events. With Peak Performance Motus software (version 8.0; Vicon, Centennial, CO), we synchronized ground reaction force and coordinate data, converted analog signals to digital signals, and filtered the coordinate data with a Butterworth filter using optimal cutoff frequencies determined via residual analysis (range, 2–5 Hz). With custom-written software (Matlab version 7.0.1; The MathWorks, Natick, MA), we performed rigid body-transformation procedures using the calibrated anatomical system technique with a single-value decomposition position and orientation estimator.14 Next, we computed clinically relevant joint angles between adjacent segments using the joint coordinate system (JCS) technique15 with positive sagittal-plane, frontal-plane, and transverse-plane rotations defined as plantar flexion, inversion, and adduction of the distal segment on the proximal segment, respectively. The exception was transverse-plane rotation of the leg segment, which was defined as medial (positive) rotation of the leg on the calcaneus.16 Trials for each participant were normalized to 100% of stance and ensemble averaged at 2% intervals. Finally, 3-dimensional displacement within 4 subphases of stance (loading response [0%–16%], midstance [16%–48%], terminal stance [48%–81%], and pre-swing [81%–100%]) was computed.17

Foot Segmentation

Foot segmentation was based on data from in vitro studies,4 in vivo roentgen stereophotogrammetric studies,46 and the concepts of constrained tarsal mechanism18 and forefoot twist.19 The model consisted of 4 functional articulations (rearfoot complex, calcaneonavicular complex, medial forefoot, and first metatarsophalangeal complex). The functional articulations and their local Cartesian coordinate systems are outlined in this subsection, and Cobb et al8 reported the details of the local reference system computation, reliability of the multisegment foot model, and agreement in the kinematic results between the multisegment foot model and invasive in vivo gait.

Rearfoot Complex. Cartesian coordinate systems defined within the leg and calcaneus segments composed the rearfoot complex (Figure 1). The JCS used to compute sagittal- and frontal-plane rearfoot complex motions was formed by the mediolateral axis of the leg segment, the anteroposterior axis of the calcaneal segment, and a floating axis computed as the cross-product of the calcaneal anteroposterior and leg mediolateral axes. To compute transverse-plane rotation of the leg with respect to the calcaneus, we constructed a separate JCS using the mediolateral axis of the calcaneal segment, the vertical axis of the leg, and a floating axis computed as the cross-product of the calcaneal mediolateral and leg vertical axes. Transverse-plane rotation of the leg relative to the calcaneus then was computed about the vertical axis of the leg.16

View Figure

• Download as Powerpoint
• Download Figure

Calcaneonavicular Complex. Cartesian coordinate systems defined within the calcaneus and the navicular segments formed the calcaneonavicular complex (Figure 1). The JCS used to compute 3-dimensional calcaneonavicular complex movements was formed by the mediolateral axis of the calcaneus segment, the anteroposterior axis of the navicular segment, and a floating axis computed as the cross-product of the navicular anteroposterior and calcaneal mediolateral axes.

Medial Forefoot. The medial forefoot was formed by Cartesian coordinate systems defined within the medial 2 rays19 and the navicular segment (Figure 1). The JCS used to compute 3-dimensional medial-forefoot motion was formed by the mediolateral axis of the navicular segment, the anteroposterior axis of the medial ray segment, and a floating axis computed as the cross-product of the anteroposterior and navicular medio-lateral axes of the medial rays.

First Metatarsophalangeal Complex. The first metatarsophalangeal complex (1MTP) was formed by Cartesian coordinate systems defined within the hallux and medial ray segments (Figure 1). The JCS used to compute 3-dimensional motions of the 1MTP was formed by the mediolateral axis of the medial ray segment, the anteroposterior axis of the hallux segment, and a floating axis computed as the cross-product of the hallux anteroposterior and 1MTP mediolateral axes.

Custom-Molded Foot Orthoses

We used a balanced traditional orthosis (BALORT) constructed with rearfoot and forefoot posting (Foot Levelers, Inc, Roanoke, VA) and a full-contact orthosis (FCORT) that provided support through the medial longitudinal arch with no rear-foot or forefoot posting (Sole Supports, Inc, Lyles, TN) (Figure 2). We chose the 2 orthoses because, although both are designed to correct abnormal foot mechanics, the methods used to affect foot function are very different. The casting procedure for both orthoses involved capturing an impression of the participant's feet in a foam box. To create a cast for the BALORT, we instructed the participant to step into the foam box. Casts for the FCORT were created using the casting procedure of the maximum arch subtalar stabilization position theory of the manufacturer. With the participant seated, we positioned the foot in a foam box. Next, we captured the impression of the participant's foot by pressing down on the thigh, along the lateral border of the foot, on all 5 toes, and on all 5 metatarsal heads. After completing the casting procedures, we sent the impressions to the manufacturers, who constructed custom-molded orthoses from the casts. The shells of both orthoses extended from the calcaneus to the level of the metatarsal heads and were covered with a full-length vinyl topcover (Figure 2). Upon receipt of the orthoses from the manufacturers, we randomly assigned participants to either the BALORT or FCORT group and provided break-in instructions. Participants were provided with only the assigned FO at the beginning of the study, so they were blinded to differences between the orthoses. We instructed participants to gradually increase the time during which they wore their FOs until they could wear them comfortably for a continuous 8-hour period. After a 2-week break-in period, the participants reported to the university's Biomechanics and Ergonomics Laboratory for gait assessment.

View Figure

• Download as Powerpoint
• Download Figure

Procedures

Before data collection, we performed dynamic camera calibration (volume = 0.5 m × 0.4 m × 0.9 m). We then applied technical marker clusters and anatomical landmarks to each segment on the participant's right foot and leg and performed an anatomical calibration procedure (Figure 1). All participants wore the same model sandal (Merrell Waterfall; Wolverine World Wide, Inc, Rockford, MI) for testing with and without orthoses. The FOs were secured to the sandals using double-sided carpet tape to prevent slippage during the orthosis condition trials. During the anatomical calibration procedure, the participant was seated with the leg oriented vertically and the midpoint of the calcaneus and second metatarsal aligned parallel to the direction of progression. We chose a semi–weight-bearing reference position because in a weight-bearing position, compensatory motions of the foot and leg already have occurred, so differences between the foot posture groups might be masked. Segmental angles computed during the no-FO anatomical calibration procedure were used as zero reference angles for the dynamic trials. After the anatomical calibration procedure, anatomical landmarks were removed, and participants performed 5 successful walking trials across a 10-m walkway at a speed of 1.3 to 1.4 m/s. We monitored walking speed using a handheld digital timer and defined a successful trial as one in which walking speed was within the appropriate range and right-limb initial contact and toe-off occurred on the force platform. Because of marker dropout during some trials, we could not reconstruct 5 trials for all participants. Therefore, 3 trials were averaged for subsequent analysis. For participants with 5 complete trials, we selected the 3 trials with the least number of marker dropouts.

Statistical Analysis

We performed repeated-measures multivariate analyses of variance (MANOVAs) for each of the functional articulations during the loading response, midstance, terminal stance, and preswing subphases. The between-groups factor in the repeated-measures MANOVAs was insert (BALORT, FCORT), and the within-group factor was insert condition (no FO, FO). Dependent variables were plantar-flexion, dorsiflexion, inversion, eversion, abduction, and adduction displacements within each subphase for each functional articulation. We computed displacement in each direction within a plane (ie, plantar-flexion and dorsiflexion displacements were computed in the sagittal plane). Follow-up repeated-measures analyses of variance (ANOVAs) were performed to investigate repeated-measures MANOVA omnibus F ratios that were different. The α level for all analyses was set at .05 (version 15.0; SPSS Inc, Chicago, IL). In addition, we computed partial η² to facilitate interpretation of the clinical meaningfulness of the results. The partial η² was interpreted based on recommendations by Cohen20 for small (0.01), medium (0.06), and large (0.14) effects. We analyzed gait kinematics within the subphases of stance using definitions of loading response, midstance, terminal stance, and preswing established by Perry and Burnfield.17

Loading Response

Repeated-measures MANOVA results did not reveal insert between-groups main effects for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 1.25, P = .36, Wilks Λ = 0.55), calcaneonavicular complex (F_6,9 = 0.86, P = .56, Wilks Λ = 0.64), medial forefoot (F_6,9 = 0.16, P = .98, Wilks Λ = 0.90), or 1MTP (F_6,9 = 0.93, P = .52, Wilks Λ = 0.62). We also did not find insert condition within-group main effects for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 0.67, P = .68, Wilks Λ = 0.69), calcaneonavicular complex (F_6,9 = 1.30, P = .35, Wilks Λ = 0.54), medial forefoot (F_6,9 = 0.98, P = .49, Wilks Λ = 0.61), or 1MTP (F_6,9 = 0.89, P = .54, Wilks Λ = 0.63). Finally, we did not find insert-by-insert condition interactions for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 1.34, P = .33, Wilks Λ = 0.53), calcaneonavicular complex (F_6,9 = 1.13, P = .42, Wilks Λ = 0.57), medial forefoot (F_6,9 = 0.57, P = .75, Wilks Λ = 0.72), or 1MTP (F_6,9 = 1.50, P = .28, Wilks Λ = 0.50).

Midstance

Midstance repeated-measures MANOVA results revealed an insert condition within-group main effect for rearfoot complex displacement (F_6,9 = 4.71, P = .02, Wilks Λ = 0.24). Follow-up repeated-measures ANOVA analysis revealed that dorsiflexion displacement was greater during the FO than the no-FO condition (F_1,14 = 5.24, P = .04, partial η² = .27) (Table 2). Participants entered midstance in a plantar-flexed position with the rearfoot complex dorsiflexing during both the no-FO and FO conditions. Upon entering midstance, participants continued to dorsiflex in a similar pattern through the entire subphase during both insert conditions (Figure 3). Although the patterns of motion were similar, participants entered midstance in a greater plantar-flexed position and transitioned from a plantar-flexed to a dorsiflexed position later during the no-FO condition (approximately 40% stance) than the FO condition (approximately 30% stance) (Figure 3). We found no insert between-groups main effects for the functional articulations of the rearfoot complex (F_6,9 = 0.89, P = .54, Wilks Λ = 0.63), calcaneonavicular complex (F_6,9 = 0.61, P = .72, Wilks Λ = 0.71), medial fore-foot (F_6,9 = 1.52, P = .27, Wilks Λ = 0.50), or 1MTP (F_6,9 = 1.63, P = .25, Wilks Λ = 0.48). In addition, we found no insert condition within-group main effects for the functional articulations of the calcaneonavicular complex (F_6,9 = 1.05, P = .45, Wilks Λ = 0.59), medial forefoot (F_6,9 = 2.03, P = .16, Wilks Λ = 0.43), or 1MTP (F_6,9 = 3.04, P = .07, Wilks Λ = 0.33). Finally, we did not find insert-by-insert condition interactions for the functional articulations of the rearfoot complex (F_6,9 = 2.60, P = .1, Wilks Λ = 0.37), calcaneonavicular complex (F_6,9 = 1.67, P = .24, Wilks Λ = 0.47), medial forefoot (F_6,9 = 1.67, P = .24, Wilks Λ = 0.47), and 1MTP (F_6,9 = 1.08, P = .44, Wilks Λ = 0.58).

View Figure

• Download as Powerpoint
• Download Figure

Table 2. Functional Articulation Excursion (°) During No Foot-Orthosis and Foot-Orthosis Conditions (Mean ± SD [95% CI])

View Fullscreen

Terminal Stance

Repeated-measures MANOVA results for terminal stance revealed insert-by-insert condition interactions for 1MTP (F_6,9 = 6.34, P = .007, Wilks Λ = 0.19) and rearfoot complex (F_6,9 = 3.75, P = .04, Wilks Λ = 0.29) displacement. Follow-up repeated-measures ANOVA analysis revealed an insert-by-insert condition interaction for abduction displacement of the 1MTP (F_1,14 = 7.87, P = .01, partial η² = .36). However, additional follow-up analyses did not reveal differences between the no-FO and FO conditions for the BALORT (F_1,7 = 4.32, P = .08, partial η² = .38) or FCORT (F_1,7 = 4.10, P = .08, partial η² = .37) (Table 2).

Follow-up repeated-measures ANOVA revealed an insert-by-insert condition interaction for eversion displacement of the rearfoot (F_1,14 = 6.64, P = .02, partial η² = .67). However, additional follow-up analysis did not reveal differences between the no-FO and FO conditions for the BALORT (F_1,7 = 3.31, P = .11, partial η² = .32) or FCORT (F_1,7 = 4.51, P = .07, partial η² = .39) (Table 2).

We found an insert condition within-group main effect for the 1MTP (F_6,9 = 6.27, P = .008, Wilks Λ = 0.19), but the difference was not investigated further because of the insert-by-insert condition interaction. We did not find insert between-groups main effects for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 1.75, P = .21, Wilks Λ = 0.46), calcaneonavicular complex (F_6,9 = 0.77, P = .61, Wilks Λ = 0.66), medial forefoot (F_6,9 = 0.67, P = .68, Wilks Λ = 0.69), or 1MTP (F_6,9 = 1.53, P = .27, Wilks Λ = 0.50). We found no insert-condition within-group main effects for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 2.81, P = .08, Wilks Λ = 0.35), calcaneonavicular complex (F_6,9 = 3.17, P = .06, Wilks Λ = 0.32), or medial forefoot (F_6,9 = 3.14, P = .06, Wilks Λ = 0.32). Finally, we did not find insert-by-insert condition interactions for the variables within the functional articulations of the calcaneonavicular complex (F_6,9 = 2.73, P = .09, Wilks Λ = 0.36) or medial forefoot (F_6,9 = 3.29, P = .053, Wilks Λ = 0.31).

Preswing

Repeated-measures MANOVA results did not reveal insert between-groups main effects for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 2.12, P = .15, Wilks Λ = 0.41), calcaneonavicular complex (F_6,9 = 0.17, P = .98, Wilks Λ = 0.90), medial forefoot (F_6,9 = 0.37, P = .88, Wilks Λ = 0.80), or 1MTP (F_6,9 = 0.62, P = .71, Wilks Λ = 0.71). We also found no insert condition within-group main effects for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 1.68, P = .23, Wilks Λ = 0.47), calcaneonavicular complex (F_6,9 = 1.99, P = .17, Wilks Λ = 0.43), medial forefoot (F_6,9 = 1.42, P = .31, Wilks Λ = 0.51), or 1MTP (F_6,9 = 1.25, P = .37, Wilks Λ = 0.55). Finally, we did not find insert-by-insert condition interactions for any of the variables within the functional articulations of the rearfoot complex (F_6,9 = 0.46, P = .82, Wilks Λ = 0.77), calcaneonavicular complex (F_6,9 = 0.56, P = .75, Wilks Λ = 0.73), medial forefoot (F_6,9 = 0.23, P = .96, Wilks Λ = 0.87), or 1MTP (F_6,9 = 1.83, P = .20, Wilks Λ = 0.45).

Limitations

Before conclusions are drawn about the effect of FO intervention for people with low-mobile foot posture, the limitations of our study should be considered. First, the changes associated with the FO intervention in participants with low-mobile foot posture were assumed to be corrective because of the similarity in the kinematic patterns between the low-mobile foot posture group in the orthosis condition in our study and previously collected data from participants with typical foot posture. However, because the participants with low-mobile foot posture in our study were asymptomatic, we could not determine whether long-term use of the orthosis would prevent or potentially contribute to the development of lower extremity injury. To further elucidate the effect of the mechanical changes associated with orthosis intervention, orthosis intervention in participants with abnormal foot posture and symptomatic lower extremity pathologic conditions or long-term orthosis intervention in asymptomatic participants with abnormal foot posture should be studied.

A second potential limitation to consider was the performance of the no-FO condition trials after the 2-week break-in period. We assumed that wearing the FOs during the break-in period would not affect gait kinematics during the no-FO condition. Although we believe the assumption is reasonable, a future study comparing no-FO condition trials before and after a period of FO intervention might be warranted.

Several other factors also deserve consideration in the planning and conduction of future studies in which the effects of FO intervention on gait kinematics are investigated. First, although we quantified foot structure and mobility using a method with moderate to high intratester and intertester reliabilities, we do not know where gait kinematics might change along the continuum of structure or mobility. Furthermore, we do not know whether foot structure, mobility, or potentially a combination thereof is related more strongly to gait function. To answer these questions, researchers need to investigate the relationship between the foot-posture measures and multisegment foot-model kinematics. Second, the influence of other factors, such as the strength of the lower extremity musculature acting as dynamic stabilizers, warrants additional investigation. Third, the effect of foot posture might become more apparent and important in situations when the lower extremity dynamic stabilizers are compromised (ie, fatigued). Fourth, although our approach partitioned stance into subphases, the statistical model continues to rely on few discrete variables to represent gait function. Future researchers also should investigate alternative approaches, such as dynamic system techniques, that might better capture the continuous nature of gait.