Foot kinematics in walking on a level surface and on stairs in patients with hallux rigidus before and after cheilectomy

All trials were performed barefoot using the marker setup and protocol according to the Heidelberg foot measurement method (HFMM) [14]. Seventeen retro-reflective markers 6 mm in diameter were attached to the skin on each leg (Figure 1), namely, on the distal phalanx of the hallux (HLX), the metatarsal heads (DMT1, DMT2, and DMT5), proximally at the 1st and 5th metatarsal (PMT1 and PMT5), the navicular (NAV), the medial and lateral malleolus (MML and LML), and the dorsal position of the calcaneus (CCL), each placed with the participant in a standing posture. The medial and lateral heel markers (MCL, LCL) were placed with the aid of an alignment device while the participant was sitting and the foot was not bearing any load. Five markers were placed at the tibia (LEP, MEP: lat./med. epicondyle, TTU: tibial tuberosity, SH1/2: two points on the medial side of the shin).

Marker trajectories were captured at 120 Hz with a 12-camera system (Vicon 612, Oxford-Metrics). The optical accuracy given by the residuum of the marker reconstruction algorithm was between 1 and 2 mm [14].

A static reference measurement was performed in a standing posture before the participant was asked to walk along a 7-m path at a self-selected speed. Data acquisition was repeated until eight full strides had been captured for each leg.

At least six stair ascents and descents were monitored on a custom-made 80-cm-wide staircase which consisted of five steps of 15 cm in height and a step distance of 32 cm.

For data processing, we only included one foot of the one patient who was operated on both feet and only one foot of each control participant (randomly chosen).

Intersegment and joint angles were calculated with the custom-made software "MoMo" within Matlab (v6.5.1) following the method described by Simon et al. [14]. This software also served to normalize the time data to the gait cycle, to average trials for each participant, to visualize charts, and to make temporospatial calculations. For further data analysis motion data were not filtered in any way since after averaging across at least 8 trials (6 for stairs) only little higher frequency content was found. Unlike typical models, which assume a series of two or three rigid and rather artificial segments in the foot and allow artificial joints of typically 3 degrees of motion (rotations) as represented by Euler- or Euler-Cardan angles, the HFMM describes the angular orientations of anatomical landmarks, possibly spanning more than one anatomical joint, rather than relying solely on rigid segment modeling. Such a "functional segment" is then described by its relative motion via projection angles defined as the angle between two vectors (or 2D-segments) in the perspective view along the axis of rotation. Consequently, the motion of the ankle complex is described by two axes of rotation accounting for talocrural and subtalar motion via the motion of the three calcaneal markers (CCL, LCL, MCL) and the navicular marker (NAV) with respect to tibial markers (LEP, TTU, SH1, SH2). For ease of interpretation, the medial arch is defined directly as the angle spanned by the triangle of the markers MCL, NAV, and PMT1. For further details concerning the model, we refer the reader to Simon et al. [14]. The reliability (between stride, rater, and day) of the HFMM parameters has already been tested in a previous study [14]. The reliability values for the hallux dorsi-/plantarflexion ROM were mean standard deviations (SD) and coefficients of multiple correlation (CMC) stride-to-stride: SD 1.37, CMC 0.993; day-to-day: SD 1.97, CMC 0.984; and inter-rater: SD 2.80, CMC 0.970. According to Simon et al. [14] the HFMM parameters show standard deviations between two and seven degrees for absolute angular values due to inaccurate marker placement on the part of the examiner but standard deviations in ROM remain small, at between 0.3° and 1.8°.

The ROM for the following parameters was chosen in order to describe the gait kinematics in the three different walking conditions (level, up stairs, and down stairs): hallux dorsi-/plantarflexion, hallux ab-/adduction, medial (longitudinal) arch (Medial arch in Table 1), subtalar in-/eversion (Subtalar motion), medial/lateral tilt of the medial arch (Medial arch inclination), midfoot-ankle pro-/supination, fore-midfoot pro-/supination, fore-hindfoot ab-/adduction, the absolute angle between metatarsal I and V, projected into the transverse plane (Metatarsal I-V angle), dorsi-/plantarflexion in the tibio-talar joint (Talocrural motion (HFMM) in Table 1) [14], and conventional ankle motion [15]. The latter parameter describes the motion of the complete foot, including mid- and forefoot motion relative to the tibia simply as a "heel" and a "toe" marker, whereas "Talocrural motion (HFMM)" [14] explicitly describes talocrural motion of the hindfoot, which is referred to as "ankle dorsi-/plantarflexion" or "ankle joint range of motion" in this paper.

For hallux dorsi-/plantarflexion, the ROM was also determined in seven functional subphases. In addition, the walking speed was determined in all gait conditions.

The statistical analyses were performed with SPSS 16.0.1 (SPSS Inc., Chicago, Ill, USA). Measures of central tendency and dispersion were calculated for all variables, and goodness of fit to normal distribution was assessed by using the Kolmogorov-Smirnov Test. The Wilcoxon paired-rank test was applied for pre-/post comparisons within the patient group; Mann–Whitney U-Test was used for group comparisons to normal references. For comparisons between conditions (level walking, stairs up and down), the Friedman test was used in order to determine overall differences. The significance level was assigned at 5% for all comparisons. Bonferroni correction was used in analyzing the seven level gait subphases. Effect sizes (Cohen’s d, effect size r) were calculated for all significant results. There is a significant difference in age (p = 0.032) and BMI (p = 0.013) between groups, which may influence the results. Therefore, regression analyses with respect to age and BMI were performed. The correlations were not significant or clinically relevant when including both patients and control participants.

Preoperatively, the pain level (AOFAS) was 18.8 ± 13.6 points (mean ± SD) and postoperatively 27.5 ± 7.1 points (0 points = severe pain, almost always present; 20 points = moderate, daily pain; 30 points = mild, occasional pain; and 40 points = no pain); according to the numbers available, no significant difference could be detected (p = 0.102). One patient was free of pain; four patients reported "mild" and three "moderate" pain. The activity level (AOFAS) was 5.9 ± 2.2 points preoperatively and 7.0 ± 2.7 points postoperatively (p = 0.083).

The preoperative limitation of the MTP-I ROM, assessed by using the AOFAS Scale, was severe in two of the eight patients, moderate in five, and mild in one. It improved to the next-better category in three cases and did not change in the others.

The mean ROM of the foot parameters in patients and control participants in all gait conditions and the results of the comparisons between groups and between time points are shown in Table 1 (means, SD and p-values).

The hallux dorsi-/plantarflexion ROM was significantly lower than in controls in level walking and descending stairs pre- and postoperatively. In the comparison between preoperative and postoperative state, the hallux dorsi-/plantarflexion ROM decreased by 2.5 degrees (p = 0.036) in level walking. Diagrams of the hallux ROM of all patients and control participants in all walking conditions are displayed in Figure 2.

The analysis of level gait subphases only showed postoperatively significant differences between patients and controls for the maximum hallux dorsiflexion in preswing (controls: 38.2 ± 5.8 degrees; patients preoperatively: 29.6 ± 6.0 degrees; p = 0.013, non-significant. after Bonferroni correction with the adjusted alpha: 0.007, postoperatively: 26.6 ± 7.1, p = 0.004). The preswing dorsiflexion did not improve after the operation (p = 0.025, non-significant after Bonferroni correction with the adjusted alpha: 0.007).

It was higher for walking down stairs than up stairs: in the control group (p = 0.033) and in the patients postoperatively (p = 0.043).

In the control group, the talocrural motion was also higher for walking stairs than for level walking (up: p = 0.013 and down: p = 0.003 versus level walking).

In both groups, the talocrural motion was higher for walking down stairs than up stairs (controls: p = 0.003, patients: p = 0.018).

Pre- and postoperative walking speeds matched in the patient group in level walking and in walking up the stairs (Table 2). Postoperatively, patients reduced their speed when walking down the stairs as compared to the preoperative speed (p = 0.043, Wilcoxon).

For patients, walking speeds were lower than for controls in level walking (Mann–Whitney U, preoperatively p = 0.034, postoperatively p = 0.021) and walking down the stairs postoperatively (p = 0.004; preoperatively p = 0.221), but not climbing up the stairs (preoperatively p = 0.329, postoperatively p = 0.242).

Cheilectomy did not restore normal hallux or ankle joint ROM in the patients of this pilot study.

The study of Nawoczenski et al. [7] found an average of 12 degrees motion recovery following cheilectomy. These authors used the proximal phalanx of the hallux instead of the distal phalanx and Cardan/Euler angle. Others [8] did not observe any significant improvements in ROM. After surgery, hallux dorsiflexion remained reduced in loading response and initial swing in those patients, but it improved during the rest of the gait phases. In our patient population, we found smaller hallux dorsiflexion in the preswing phase of gait postoperatively, indicating a slight deterioration rather than an improvement. These patients might try to secure the more mobile MTP-I joint in order to prevent pain.

Reasons other than bony abutment seem to account for the residual limitation in hallux mobility. This finding could be related to capsular shrinkage or scar tissue in case of delayed physiotherapy or mobilization after the operation. Other reasons for the small functional effects observed for cheilectomy could be due to concomitant pathological conditions such as stenosing tenosynovitis of the flexor hallucis longus, which were not assessed by the bony resection without a plantar soft tissue release. According to the graphs for the hallux ROM over the gait cycle (Figure 2), the dorsiflexion curve drops steeper than in the control group in the initial swing phase. This effect could be due to a passive force, perhaps caused by a soft tissue restriction acting as a reset force. Therefore, we strongly recommend that exercising the MTP-I ROM is begun immediately in postoperative treatment. The motion pattern of a person who has walked with a limited ROM of the MTP-I joint for years might also remain the same even one year after the operation.

Furthermore, ROM was not completely restored in the horizontal plane by the operation: The hallux ab-/adduction increased after surgery, but it was still significantly lower than in controls. The osteophytes might not have been the main cause of this restriction, but rather the morphologic changes in the MTP-I joint due to osteoarthritis. However, the restricted fore-hindfoot ab-/adduction might also be a secondary consequence of the abnormal forefoot kinematics as it did not improve during level walking after the operation. Indeed, the reduced mobility in the sagittal plane seems to cause secondary restrictions in the other degrees of freedom (ab-/adduction and pro-/supination). Comparable restrictions in the mobility of adjacent foot segments in planes (parameters: tilt of the medial arch, in-/eversion) other than the initially mostly restricted one (plantar-/dorsiflexion) were previously observed in patients with ankle osteoarthritis after joint replacement [19].

The preoperative limitations in the MTP-I joint were expected to be more relevant for climbing/descending stairs than for ground level gait. Contrary to our expectations, however, the hallux ROM was the highest in level walking, followed by descending stairs. Apparently, there is no need for extended ROM of the MTP-I joint while climbing stairs; compared to level walking a significantly lower ROM was found in climbing stairs.

In previous studies, the kinematics of climbing stairs showed increased knee and hip flexion in the sagittal plane as compared to level walking in healthy participants [9‐11]. In addition, the ankle joint is more involved in the swing phase [20]. Greater ankle angles were found while descending stairs [21].

In both of our groups, the ankle joint ROM was significantly higher in walking stairs than in level walking. The higher need of mobility for the level change in walking stairs apparently does not involve the forefoot, but mostly the ankle joint. Our findings of higher ankle ROM while descending stairs than ascending stairs confirmed the results of previous studies [21].

The same was also found for the MTP-I joint in our healthy participants and for patients postoperatively. This finding could be interpreted to indicate a slight improvement towards a normal gait pattern.

In all gait conditions, we found a significantly lower fore-midfoot pro-/supination ROM in patients than in controls, with an improvement in level gait after the operation. Our results suggest that this effect was successfully treated by cheilectomy, although the ROM of the MTP-I joint did not change significantly. Postoperatively, the group difference only remained significant for descending stairs.

The reduced fore-/midfoot pro-/supination could represent one of the main adaptation mechanisms in hallux rigidus patients due to the limitation in the MTP-I joint. As rolling over the first ray is limited, the foot might stay in a more locked and supinated position in the mid- to forefoot area. Therefore, this ROM was also reduced as a secondary consequence of the pathology in the MTP-I joint. The difficulty hallux rigidus patients have in walking down steps could also be due to the additional rigidity in the midfoot. The operation might have had an additional negative influence on the sense of security for walking down stairs as patients also reduced their walking speed.

The main goal of any operative intervention to treat hallux rigidus is to reduce pain. Although there was no statistical evidence of pain relief, we found an overall improvement in the total AOFAS Scale after the operation, indicating a better clinical state. The lack of significant changes in the subcategories pain and activity could be due to the small number of patients. The validity of the AOFAS Hallux Metatarsophalangeal-Interphalangeal Scale has previously been shown to be questionable, especially related to activity [22]. Not all the subcategories are useful for specifying the pathology of hallux rigidus: Since ROM is not differentiated in dorsiflexion and plantarflexion of the MTP-I joint, restrictions and changes in the MTP-I joint dorsiflexion are not always well represented. The motion of the interphalangeal joint of the hallux and the stability of the metatarsophalangeal-interphalangeal joint are mostly not impaired by hallux rigidus. Therefore, these subcategories do not reflect the changes achieved by cheilectomy, in which osteophytes are always removed. The weighting of the alignment seems to be mostly chosen with respect to hallux valgus patients. Therefore, these subcategories are not suitable for distinguishing the degree of hallux rigidus and quantifying the results of the operative procedure. However, the total scale sum seemed to reflect the state of the patient quite well since the main points are composed of subjectively reported data.

The minor effect of the operation on foot kinematics and kinetics may be due to selecting patients with hallux rigidus grade I and II [12]. Usually, only these patients would be treated by cheilectomy; in patients with more severe disease and greater ROM limitations, arthrodesis would be performed or a hemi-prosthesis implanted. Patients with a more severe degree of osteoarthritis in the MTP-I joint would probably show greater changes in these outcome parameters after the operation.

Furthermore, the relatively low sample size needs to be mentioned. Therefore, this study should be considered as a pilot study. Prospectively, the effects of cheilectomy on hallux function in walking stairs were lower than expected. However, our data show clinically relevant differences between the pathologic and physiologic foot kinematics. Therefore, using the HFMM seems to be appropriate for investigating the pathological state of the MTP-I joint. The HFMM does not directly measure MTP-I joint motion. Rather, it measures motion of the distal phalanx with respect to the first metatarsal. However, we excluded major compensatory changes in the first interphalangeal joint (i.e., hyperextension deformities) in the thorough clinical assessment. In addition, previous studies have implicated (dorsal) translation of the first metatarsal [23‐25] that may minimize angular changes of the proximal phalanx. The HFMM might miss these kinematic responses.