Twenty-one subjects (six men and fifteen women) who had residua of unilateral congenital dysplasia of the hip were entered in this study (Table I). The mean age was forty-eight years (range, twenty-five to seventy-one years). No subject had had previous operative or non-operative treatment of the hip, and no subject had any other disability related to a bone or a joint. These subjects were compared with a control population of forty subjects (fourteen men and twenty-six women) who had no known abnormality of the locomotor apparatus (Table II). The mean age of the control subjects was forty-six years (range, thirty-one to seventy-one years). As differences in gait parameters between men and women have been described12, whenever individual subjects were analyzed the comparison was made with the control subpopulation matched by gender.
Clinical and Radiographic Evaluation
All of the subjects who had residua of congenital dysplasia of the hip were examined clinically and radiographically. Clinical assessment of both hips was performed with use of the Harris hip score. An anteroposterior radiograph of the pelvis and an axial radiograph (the so-called false-profile radiograph of Lequesne and de Seze) of both hips were made. Three parameters were measured bilaterally2 on the anteroposterior radiograph: the VCE angle, the HTE angle, and the CC'D angle (Fig. 1). The VCE angle (the Wiberg angle; normal value, 25 degrees or more) is the angle between a vertical line (V, as shown in Fig. 1) passing through the center of the femoral head (C, as shown in Fig. 1) and a line drawn from the center of the femoral head to the most external point of the acetabular roof (E, as shown in Fig. 1). The HTE angle (normal value, 10 degrees or less) is the angle between a horizontal line joining the symmetrical internal points of the right and left sides (T, as shown in Fig. 1, and T', which is not shown) of the acetabular roof and a line between the ipsilateral internal point (T) and the most external point of the acetabular roof (E). The CC'D angle (normal value, 137 degrees or less) is formed by the longitudinal axis of the femoral neck and the longitudinal axis of the femoral diaphysis.
The VCA angle (normal value, 25 degrees or more) was measured bilaterally on the axial radiograph (Fig. 1). This angle measures the anterior coverage of the femoral head, as defined by a vertical line (V) passing through the center of the femoral head (C) and a line joining the center of the femoral head and the anterior point of the acetabular roof (A)2.
Femoral anteversion was calculated with use of the technique of Magilligan2, which requires consideration of both the anteroposterior and the axial radiographs (as described by Arcellin2). The joint space (measured on the anteroposterior radiograph as the smallest distance between the femoral head and the acetabulum) and the interruption of the Shenton arch, which is an indication of the degree of subluxation of the hip, were also measured. No subject had superior dislocation of the femoral head with avascular necrosis.
Evaluation of the control subjects consisted of only anamnestic inquiry and physical examination to exclude any abnormality of the locomotor apparatus.
Analysis of Gait and Experimental Setup
The kinematic data were collected with the ELITE system5. Briefly, four video cameras make an image of the subject and send it to a dedicated processor that, by a pattern-recognition procedure, detects the passive reflective markers glued on the subject at the level of the seventh cervical vertebra, the apophysis of the eighth and ninth thoracic vertebrae, the sacrum, the posterior superior iliac spines, the lateral epicondyle at the knees, the lateral malleoli, and the fifth metatarsophalangeal joints. The coordinates are computed in real time (100 times per second). The three-dimensional coordinates of each marker are then computed, and the kinematic variables are calculated from these6,14. The skin-to-bone movement and muscle contraction at the sites of the markers were considered to have little effect on the kinematic measurements. One additional marker was placed on each side of a rigid probe connected by a lightweight bracket to the medial and lateral femoral condyles to allow analysis of the internal and external rotations of the thigh along its long axis. A local reference coordinates system was defined for the pelvis, with the origin at the mid-point between the two posterior iliac spines, the z axis parallel to a line connecting the two posterior iliac spines and directed to the right, the x axis perpendicular to the plane of the three posterior markers and directed forward, and the y axis perpendicular to the previous two axes and directed superiorly.
A force platform with twelve piezoelectric force transducers embedded in the walkway was used to measure the ground-reaction forces. The platform was connected to the computer, and the force data were processed to obtain a graphical representation of the resultant ground-reaction vectors3,13 and were also combined with kinematic data to compute the externally applied dynamic joint moments and powers15. The center of rotation of the hip joint was identified by measurement of the radiographs of each patient and was transferred, after proper scaling, to the local reference system of the pelvis. The center of the knee joint was assumed to be on the mid-point between the lateral and medial femoral condyles. The center of the tibiotarsal joint was assumed to be in the middle between the lateral and the medial malleoli (the intramalleolar line was considered to be perpendicular to the plane that is defined by the markers at the knee, ankle, and fifth metatarsal head). All of these points were computed on the basis of individual measurements made for each subject. The linear and angular coordinates were differentiated twice to obtain the velocities and accelerations of all segments of the lower limb4. The dynamic equilibrium equations were then solved in relation to the center of each joint of the lower limb1 by adopting anthropometric parameters (center of mass location, mass, and principal moments of inertia) taken from regression equations22 and scaled for each subject. The output of this procedure was the time-course of the joint angles, velocities, and moments of forces. The joint power was determined by multiplying the externally applied joint moment by the angular velocity of the joint. The stride temporal phases (stance phase, swing phase, and double-support phase) were identified bilaterally by analysis of the ground-reaction forces and the position of the feet in relation to the floor. The length of the stride was measured as the distance between two subsequent contacts of the same foot with the floor. The length of the step was measured as the distance between the two feet during the double-support phases along the line of progression. The duration of the stride, mean velocity of progression, and mean velocity of the feet (during the swing phase) were calculated from the aforementioned parameters.
Acquisition and Normalization of the Data
The measurements were made while the subjects walked barefoot on a ten-meter-long walkway. Data were collected during ten trials in which the left and right feet were alternately placed on the force platform. Measurements were recorded once the subject had achieved a steady gait. Although some of the subjects who had dysplasia used a cane to walk long distances, all measurements of gait were made while the subject walked without the use of external support. The subject chose the cadence that was most natural and comfortable. Data were also recorded during quiet standing and during left and right monopodal standing. For ten subjects who had dysplasia, data were acquired again after one month to evaluate the reproducibility of the results. For all ten subjects, the individual features of gait were reproduced in the second recording session and the significant differences from the control subjects (in terms of joint angles, moments, and powers) were confirmed.
The acquisition of data with real-time processing required about twenty minutes for each subject.
In order to compare the data, different kinds of normalization were applied. The durations of the stride temporal phases were expressed as a percentage of the duration of the whole stride. The length of the stride, the length of the step, and the mean velocity of progression were expressed as a percentage of the height of the subject.
All of the curves representing the time-course of different variables were normalized in time (the duration of the whole stride was considered as 100 per cent). Mean curves and standard deviations were then calculated. For the variables that depended on body weight (typically, the ground-reaction force components, joint moments, and joint powers), amplitude was normalized by dividing by body weight before averaging19. To compare the data from individual subjects who had dysplasia with those of the gender-matched control subjects, the mean normalized curves of the selected control subjects were multiplied by the body weight of each subject who had dysplasia. This allowed for a representation in which the original information regarding amplitude was maintained. Spatiotemporal parameters, and the maximum and minimum values in each curve (amplitude, latency, and peak-to-peak difference) were compared with the corresponding parameters for the control subjects with use of the two-tailed Student t test. The correlation with the Harris hip score was analyzed with use of the linear correlation coefficient.
Clinical and Radiographic Findings
The radiographic findings of deformity and degenerative changes of the hip were consistent with those noted during the clinical evaluation (Table III).
Eight subjects (Subjects 4 through 7, 10, 17, 19, and 21) had some degree of dysplastic alterations in the contralateral hip, which was either completely asymptomatic or slightly painful only occasionally.
Analysis of Gait
Three-Dimensional Computed Reconstruction of the Body (Stick Diagram)
A quantitative analysis was performed by projecting a stick diagram of the torso, pelvis, and lower limbs in the sagittal, frontal, and horizontal planes. The tilt angle of a line passing through the posterior iliac spines with respect to the horizontal reference plane of the laboratory was measured to study the Trendelenburg sign. The control subjects walked with a slight pelvic drop (mean and standard deviation, 2.1 ± 2.2 degrees) during the stance phase, which was significantly different from zero (p = 0.05) (Fig. 2, A). The subjects who had a dysplastic hip (Fig. 2, B and C) had a significant increase (p = 0.05) in the drop of the pelvis (mean pelvic drop, 5.3 ± 5.1 degrees) during the stance phase on the affected side, and the drop was significantly correlated with the Harris hip score (p = 0.01; r = 0.75). In contrast, during the stance phase on the unaffected side, the drop of the pelvis was either reduced, absent, or reversed, depending on the severity of the degenerative conditions. The mean value was 0.2 ± 5.0 degrees (indicating that the affected hip was higher than the unaffected one), but there was a strong correlation with the Harris hip score (r = 0.68).
Projection of the stick diagram on the horizontal plane showed that the control subjects had alternate pelvic rotations (Fig. 3, A). The angle between the mediolateral (z) axis and the perpendicular to the plane of progression in a horizontal projection ranged from 4.8 degrees (one side forward) to -4.8 degrees (the contralateral side forward) with a standard deviation of 2.4 degrees. The subjects with a dysplastic hip had reduced forward displacement of the contralateral side; in the subjects who had more severe dysplasia, the unaffected hip never surpassed the affected one (Fig. 3, B and C). When the affected side was used as a reference, the mean angles of pelvic rotation in our test population ranged from 7.8 ± 7.3 degrees to 0.4 ± 7.0 degrees, both of which values were significantly different from the normal values (p = 0.05). Although this abnormal orientation of the pelvis was more evident in the subjects who had more severe dysplasia, there was no significant correlation with the Harris hip score or with the radiographic parameters.
Stride Parameters
The difference between the duration of the stance phase on the affected side and that on the unaffected side (r = -0.63), the foot velocity on the affected side (r = 0.62), and the difference between the foot velocity on the affected side and that on the unaffected side (r = -0.67) all correlated significantly with the Harris hip score (Table IV). The correlation of the duration of the stance phase on the unaffected side (r = -0.53) was only significant at p = 0.05. The other parameters showed little or no correlation with the Harris hip score. The trend was toward a reduced gait velocity and length of stride as the functional score decreased, with the duration of the stride increasing as a consequence. Surprisingly, despite a marked asymmetry of gait, neither the double-support time nor the length of the step correlated with the Harris hip score.
Ground-Reaction Forces
In the control subjects, both the first and the second peaks of the ground-reaction force were about 110 to 120 per cent of the body weight, depending on the walking cadence (Fig. 4, A). For all but two of the subjects who had dysplasia, the first peak was reduced to less than body weight (Fig. 4, B). The second peak was reduced only in some of the most symptomatic subjects (Fig. 4, C). In the most severely affected subjects, the morphology of the ground-reaction force display was markedly abnormal (Fig. 4, D) and a plateau was observed in the rising phase. The first peak of the clinically asymptomatic limb was reduced in only four subjects who had dysplasia.
Joint Kinematics
In all subjects who had dysplasia, the affected hip showed a precocious and progressively more marked reduction of extension during the single-limb stance phase (Figs. 5-A and 5-B); in the subjects who had the most severe dysplasia (as indicated by a Harris hip score of less than 65 points), a lack of extension was observed as well as a reduction of flexion (Figs. 5-C and 5-D). The maximum range of motion and the extension peak (the minimum hip angle) both correlated with the Harris hip score (r = 0.86 and -0.74, respectively; p = 0.01).
The time-course of the flexion-extension angle of the knee showed a progressive disappearance of knee flexion at load acceptance with a lower Harris hip score and progressive reduction of maximum flexion during the swing phase (Figs. 5-A, 5-B, 5-C through 5-D). The maximum angle of knee flexion during the swing phase significantly correlated with the Harris hip score (r = 0.79; p = 0.01).
The most apparent feature at the ankle joint was a reduction of plantar flexion, compared with the value in the control subjects during early swing phase in subjects who had a low Harris hip score (Figs. 5-C and 5-D). The excursion of the ankle joint was consequently reduced in the subjects who had dysplasia. However, these parameters were not correlated with the Harris hip score. Instead, the maximum dorsiflexion of the ankle correlated with the Harris hip score (r = 0.67; p = 0.01).
Thirteen subjects with a Harris hip score of less than 75 points had permanent external rotation of the affected hip (a maximum peak that was significantly different from that in the normal controls [p = 0.05]), while the unaffected hip was internally rotated for most of the stride cycle. In the same subjects, adduction of the affected hip was greater than normal (p = 0.05), with a peak in mid-stance.
Joint Kinetics
Both the externally applied flexion moment and the extension moment about the hip during the stance phase were significantly reduced (p = 0.05) in the sagittal plane in all of the subjects who had dysplasia (Figs. 6-A, 6-B, and 6-C). In all thirteen subjects who had a Harris hip score of less than 80 points, the reduction of the extension moment about the hip was greater than the reduction of the flexion moment about the hip. In these subjects the duration of the flexion moment about the hip was also prolonged, and in the subjects who had the lowest Harris hip scores the external flexion moment lasted for most of the stance phase (Fig. 6-D). The externally applied extension moment about the hip became negligible in subjects who had a lower Harris hip score and remained essentially unchanged for all of the swing phase. There was a strong correlation (r = -0.8; p = 0.01) between the peak extension moment about the hip and the Harris hip score.
The peak flexion moment about the knee, which occurred during early stance phase, was progressively reduced toward zero as the Harris hip score decreased (Fig. 6-B). In three subjects who had a low Harris hip score, the flexion moment about the knee was reversed in direction, becoming extensory for most of the stance phase (Figs. 6-C and 6-D). The maximum applied external flexion moment about the knee correlated with the Harris hip score (r = -0.78; p = 0.01). The second peak of the externally applied flexion moment was not reduced and showed no significant correlation with the Harris hip score.
There was no significant reduction of the peak externally applied dorsiflexion moment at the ankle during the push-off phase. However, slight changes in the morphology of the curves were observed qualitatively among the different subjects who had dysplasia (Figs. 6-A, 6-B, 6-C through 6-D). The most frequent alteration was a reduction in, or lack of, plantar flexion moment during the early stance phase and a subsequent rapid increase of the dorsiflexion moment. Four subjects, who had a Harris hip score of more than 80 points, exhibited an opposite behavior, with a prolonged plantar flexion moment during early stance phase and a delayed increase of the dorsiflexion moment.
In the frontal plane, the torque about the hip joint was changed in different ways in different subjects. The ten subjects with dysplasia who had a Harris hip score of more than 60 points had an increase in the adduction moment (Fig. 7-A), with both peaks in the stance phase being significantly different (p = 0.05) from the corresponding peaks for the control subjects. Of the remaining eleven subjects, seven had a general reduction of the moment and a change of the typical time-course (Figs. 7-B and 7-C) and four unexpectedly exhibited a significantly higher-than-normal (p = 0.05) adduction moment of both peaks (Fig. 7-D). All four had important radiographic signs of degeneration of the hip; one (Subject 12) had marked atrophy of the muscles of the hip, with a severe limp. The other three subjects had the lowest Harris hip scores.
The joint moment in the frontal plane at the knee and the ankle was, in general, within the range of that of the respective normal controls; there were only minor morphological deviations (Figs. 7-A, 7-B, 7-C through 7-D).
Power
At the hip joint, the peak of power absorption during the mid-to-late stance phase and the peak of power generation during the late stance-to-early swing phase were both markedly reduced (p = 0.05) in all subjects who had dysplasia (Figs. 8-A and 8-B). Both peaks strongly correlated (r = 0.82 and r = 0.85, respectively; p =, with the Harris hip score. In the subjects who had a Harris hip score of less than 70 points, all three peaks of power at the hip joint, both generation and absorption, were reduced to close to zero. No significant reductions were observed in the unaffected hip.
At the knee joint, there was a progressive reduction of the absorbed power with reduction of the Harris hip score (r = 0.77; p = 0.01). The power of the contralateral knee and ankle joints usually was normal or slightly reduced.
The aim of the present study was to understand the progressive degeneration of the gait of subjects who have residua of congenital dysplasia of the hip. We organized our investigation to obtain the most complete information on gait in a group of subjects who had various degrees of impairment as a result of this disease.
The subjects who had dysplasia appeared to walk with a reduced velocity, short steps, long periods of stance on the unaffected side, a particular attitude of the trunk and the pelvis, and reduced amplitude of rotation of the hip joint. Some of these features have been described in previous studies8,11,21 and were well documented in our investigation through a quantification of spatiotemporal parameters and angular variables. In addition to those of previous studies, the results of our analysis provide information on variables that cannot be estimated visually, such as the ground-reaction forces, the joint moments, and the joint powers. This information is important for an interpretation of the adaptation mechanisms and for the correlation of kinematic changes with the clinical grading score.
The subjects who had dysplasia had asymmetrical oscillations of the pelvis in the frontal plane. The control subjects had slight pelvic drop during single-limb support. This was interpreted by Saunders et al. as a way to reduce the energetic requirement for progression of the body "by the cutting of the vertical displacement of the center of gravity in half." Our results clearly demonstrate that the greater the degree of degenerative change in the dysplastic hip joint, the more the pelvis tended to fix itself in an oblique position during all phases of gait so that the affected hip was always maintained higher than the unaffected one. This phenomenon is probably linked to a primary weakness of the abductor muscles or to a complex interaction of contractures of the adductor muscles and pain and is necessarily associated with spinal scoliosis, varus thrust of the dysplastic hip, and an increased lateral displacement of the pelvis during walking (to balance the weight of the body). This postural attitude and these deformities augment the insufficiency of the acetabulum and the apparent limb-length discrepancy. The pelvis was asymmetrically rotated in the horizontal plane to maintain the affected side in front of the unaffected side. The rotation of the pelvis, with the foot correctly oriented and fixed on the floor during the stance phase (Fig. 3), resulted in a prolonged external rotation of the affected hip and in internal rotation of the unaffected hip. The fixed external rotation of the dysplastic hip undoubtedly increased the anterior subluxation of the femoral head, increasing the stress on the anterior part of the acetabulum, which usually already had a reduced surface area.
The range of motion of the hip, particularly extension, was reduced. Extension was decreased in subjects who had a history of early disease and was more marked in subjects who had progressive degenerative changes. As a consequence of the lack of extension of the hip, the knee compensated with flexion to allow the pelvis to progress forward before toe-off.
The ground-reaction forces associated with the weight-acceptance phase were reduced for all of the subjects who had a dysplastic hip. Only the more severely affected subjects had a reduced push-off force for the affected limb. These changes can be interpreted as an adaptation to reduce the load on the joint and, thus, as a protective compensation.
The moments of forces around the hip were usually reduced in the sagittal plane. This reduction represents a protective mechanism adopted by the neuromuscular system to reduce the loads on the abnormal joint. This adaptation is probably related to the alteration of mobility of the hip joint and to the proprioceptive and nociceptive input.
Different moments in the frontal plane correlated with different clinical and functional conditions. The slightly symptomatic subjects had an increased adduction moment about the hip. We are not able to explain this finding or to determine whether the increase of the adduction moment is the result or the cause of the progression of degenerative alterations associated with dysplasia of the hip. It is tempting to say that an increased adduction moment, linked to an inadequate neuromuscular response or to a modified proprioceptive input, could have been one of the causes of pain in the subjects who had dysplasia.
The adduction moments about the hip tended to decrease in the subjects who had more severe dysplasia. This is more easily explained as the result of a strategy of the nervous system to reduce pain and to save the joint.
Four of the subjects who had more severe dysplasia had an increased adduction moment about the hip. As before, this increase could be associated with poor proprioceptive adaptive mechanisms that favor the worsening of clinical conditions. These subjects could be at a higher risk of precocious deterioration of the hip joint, and the increased adduction moment could be a risk factor for a poor clinical outcome after an operation, similar to that shown for the knee joint16.
Hip power was reduced or abolished in all subjects who had dysplasia. This could be partially due to the decrease in walking velocity and the reduction of the forces exerted around the hip joint. It is of note that the power of the ipsilateral knee and ankle joints as well as of the contralateral hip were not changed much, and thus reduction of gait velocity was not sufficient to explain the phenomenon.
The three-dimensional position of the pelvis during gait, with the affected hip higher and anterior with respect to the unaffected hip, in the subjects who had dysplasia of the hip appeared even in early stages of the disease and before any substantial degenerative change had taken place. It should be noted that subjects who had a Harris hip score of more than 80 points did not have any clinically detectable reduction of mobility of the hip joint. We believe that every preoperative and postoperative rehabilitation program should take into account this abnormal position of the pelvis, which increases the detrimental effect of an insufficient acetabulum.
We do not know how operative treatment, such as femoral or pelvic osteotomy and hip replacement, affects the abnormal position of the pelvis. We believe that these dynamic pelvic abnormalities affect muscle forces, changing the length and direction of the active muscles, and are responsible for an altered distribution of loads at the joint. Thus, the results of clinical treatment need to be investigated.
The most important factor producing modifications of gait was probably pain and altered proprioceptive input; similar results were reported10 in patients who had coxalgia secondary to osteoarthrosis or necrosis of the femoral head. Our results show that even subjects who did not have pain during the test had modified their gait pattern. Therefore, it is likely that experience was sufficient to make the subject develop an antalgic gait pattern. Concerning the moments in the frontal plane, we can only speculate as to why some subjects adopted compensating mechanisms to reduce the adduction moment while others did not. An increased adduction moment about the hip in the most severely affected subjects can be interpreted as the late result of the natural evolution of the abnormality, a stage at which the subject no longer is capable of maintaining the protective compensating mechanism. Reasons for this can be the progressively increasing weakness of the abductor muscles or the adductor contractures, or both, that take place in the advanced stages of coxarthrosis of the hip. As a consequence, increasingly pronounced lateral displacement of the pelvis would be necessary to balance the drop of the pelvis. This, in turn, could increase the lever arm of the ground-reaction force at the affected hip, the required adduction moment, and consequently the stress on the hip. The proper time for operative treatment could be before the occurrence of instability, as revealed by an increase of the adduction moment about the hip. Also, a preoperative rehabilitation program aimed at recovery of the hip muscles and correction of the instability of the pelvis could be devised on the basis of gait-analysis data. A prospective study would help to verify this hypothesis.
NOTE: The authors thank Mr. Mauro Recalcati, Mr. Michele dalla Mura, and Mr. Alessandro Santagostini for their valuable help in the collection and elaboration of the data.