 |
 |
 |
|
| home | CME | help | feedback | subscriptions | archive |
|
© 1997 The Journal of Bone and Joint Surgery, Inc. The Effect of Limb-Length Discrepancy on Gait*   , DALLAS, TEXASInvestigation performed at Texas Scottish Rite Hospital for Children, Dallas
We evaluated the gait of thirty-five neurologically normal children who had a limb-length discrepancy of the lower extremities that ranged from 0.8 to 15.8 per cent of the length of the long extremity (0.6 to 11.1 centimeters). The twenty-two boys and thirteen girls had an average age of thirteen years (range, eight to seventeen years). No patient had a substantial angular or rotational deformity of the lower extremities. We found no correlation between the actual discrepancy or the per cent discrepancy and any of the dependent kinematic or kinetic variables, including pelvic obliquity. Discrepancies of less than 3 per cent of the length of the long extremity were not associated with compensatory strategies. When a discrepancy was 5.5 per cent or more, more mechanical work was performed by the long extremity and there was a greater vertical displacement of the center of body mass. Clinically, this degree of discrepancy was manifested by the use of toe-walking as a compensatory strategy. Children who had less of a discrepancy were able to use a combination of compensatory strategies to normalize the mechanical work performed by the lower extremities.
Some authors have recommended that a limb-length discrepancy of two centimeters or more at the end of osseous growth should be treated operatively3,27,35; however, we were unable to find a well designed study to support this recommendation. Several reports have suggested that inequality in the lengths of the lower extremities is associated with back pain10-12,24,26,29,32, dysfunction of the knee5,23,25, and osteoarthrosis of the hip10,13,14,26. We found only one study regarding the biomechanical alterations in gait secondary to an acquired limb-length discrepancy22. We evaluated the effect of limb-length discrepancy on the gait of otherwise healthy children. We documented the strategies used to compensate for large discrepancies, and we also attempted to identify a threshold of discrepancy above which there were biomechanical alterations in gait.
We reviewed the records of 275 patients who had been evaluated at our institution, between 1985 and 1993, for a limb-length discrepancy. Approval for this study was obtained from the Institutional Review Board. The criteria for inclusion were a minimum age of seven years (to ensure that the child had established a mature pattern of walking34); a clinically detectable limb-length discrepancy; less than 10 degrees of angular difference between the two extremities in any plane; less than 10 degrees of difference in the total range of motion of the hips, knees, and ankles in any plane; and normal neurological and cognitive function. Children who had had an operation to equalize the lengths of the extremities within the preceding six months, had had an injury of the lower extremity that had necessitated immobilization or the use of crutches within the previous six months, or had had pain in the lower extremity or back that restricted their activity were excluded from the study. Forty-three children met our criteria for inclusion, and thirty-five of these children and their families agreed to return to our hospital for evaluation. There were twenty-two boys and thirteen girls, with an average age of thirteen years (range, eight to seventeen years). Fourteen patients had a short left extremity and twenty-one had a short right extremity. The etiology of the limb-length discrepancy was idiopathic in nine patients, a congenitally short femur in six, fibular hemimelia in five, congenital hemihypertrophy in five, tibial bowing in four, fracture of the femur in three, Aitken type-A proximal femoral focal deficiency1 in three, and fracture of the tibia in one (Fig. 1). An additional two children had had closed reduction of a congenital dislocation of the hip, after which a limb-length discrepancy developed without radiographic evidence of avascular necrosis of the hip or acetabular dysplasia. Three patients had two diagnoses each. Nine children had had an operative procedure to correct the discrepancy: six had had an epiphysiodesis of the distal aspect of the femur as well as of the proximal aspect of the tibia and fibula more than one year before the study, and three had had a femoral or tibial lengthening with removal of the frame at least eight months before the study.
We selected patients who had a limb-length discrepancy without associated neurological, structural, or muscular abnormalities that could alter the biomechanics of gait. Initially, we had intended to include only patients who had less than a 15 per cent difference in maximum torque per body weight between each of the muscle groups in the long extremity and the corresponding group in the short extremity. This degree of variability was considered acceptable on the basis of reports that such a difference is seen in otherwise healthy individuals2,15,19. However, all of our patients had more than this degree of variability between at least one pair of muscle groups. The differences could not be correlated with the underlying diagnosis, degree of discrepancy, primary site of discrepancy (tibia or femur), or compensatory strategies. We could not determine whether these differences were due to the underlying etiology of the limb-length discrepancy or were secondary adaptations to the discrepancy. All patients had a physical examination, which included evaluation of the range of motion and the presence of angular deformity, instrumented gait analysis, and isokinetic testing of the muscles of the lower extremities. An orthoroentgenogram16 was made if one had not been made within the previous four months. The limb-length discrepancy was measured with use of blocks placed beneath the short limb, a tape measure from the anterior superior iliac spine to the medial malleolus4,9,28, surface markers used in gait analysis, and the orthoroentgenogram. The limb-length discrepancy was expressed as the absolute difference in centimeters as measured on the orthoroentgenogram and as a percentage of the length of the long extremity. The children walked at a self-selected speed along a fifteen-meter runway that included a calibrated space of 2.5 meters. A videotape of each walk was made simultaneously from the frontal and sagittal planes. Reflective markers were placed on the patient at several anatomical landmarks: the base of the sacrum midway between the posterior superior iliac spines, both anterior superior iliac spines, the lateral epicondylar ridge of the distal end of the femur along the flexion-extension axis of the knee, the lateral aspect of the thigh along the axis of the knee (on a ten-centimeter-long aluminum wand fastened here), the most prominent point of the lateral malleolus along the transmalleolar axis, in line with the transmalleolar axis (on an aluminum wand), the midfoot between and slightly proximal to the second and third metatarsal heads, the posterior aspect of the glenohumeral joint, and the radial styloid process. Three-dimensional kinematic data were recorded with a six-camera sixty-hertz VICON system (Oxford Metrics, Oxford, England). Kinetic data were recorded with two force-plates instrumented with strain-gauges (AMTI, Newton, Massachusetts). Four force-plate strikes were recorded for each extremity for each patient. Joint angles; internal moments; and powers in the sagittal, transverse, and coronal planes were calculated with VICON Clinical Manager software (Oxford Metrics). These values were calculated separately for each extremity since the software averages the lengths of the limbs as a scaling parameter when the location of the center of the hip joint is determined. Cadence, velocity, step length, and percentage of gait cycle in single and double-limb stance were also recorded. The total mechanical work performed by each lower extremity was determined by integrating and adding the absolute values of the areas under the hip, knee, and ankle power curves in all planes8. We calculated a position for the total center of body mass segmentally in the x, y, and z directions for each trial, and determined the displacement of the center of body mass during the gait cycle, on the basis of the assumptions that each upper extremity was 5 per cent of the total body mass, the head and trunk were 58 per cent, each thigh was 10 per cent, and each shank and foot was 6 per cent. We calculated the rate of loading at heel-contact from the vertical z force of the force-plate for each trial. Heel-contact velocity and acceleration were estimated from the displacement of the heel marker for each trial, and a coefficient of variation for the four trials for each extremity was determined. A Cybex II machine (Lumex, Bayshore, New York) was used to test isokinetically the muscle strength of the hip flexors and extensors with the patient in the supine position, the plantar flexors and dorsiflexors of the ankle with the patient in the prone position, the abductors and adductors of the hip with the patient in the side-lying position, and the flexors and extensors of the knee with the patient seated. The hip and ankle were tested at 30 degrees per second, and the knee was tested at 60 degrees per second. The peak torque per body weight was selected from one of five trials. We calculated the Spearman non-parametric and Pearson correlation coefficients between the absolute or per cent limb-length discrepancy and the difference between the long and short limbs with regard to the dependent kinematic and kinetic variables for the pelvis, hip, knee, and ankle. Spearman coefficients of more than 0.8 were considered clinically predictive. With use of the Pearson correlation coefficient, we looked for any relationship between the per cent limb-length discrepancy and the dependent kinetic and kinematic variables in the twenty-eight children who had a discrepancy of less than 6 per cent and in the seven who had a discrepancy of 6 per cent or greater.
Classification of Compensatory Strategies Paired t tests were used to compare the coefficients of variation of the long extremity with those of the short extremity to determine whether the variables were more consistent for one extremity. Two-sample t tests were used to determine if there were any differences between the long and short extremities with regard to dependent gait and Cybex variables that would provide objective criteria with which to define the compensatory strategies. Children who were visually classified as using toe-walking did not have the normal first and second ankle rockers described by Perry30. All of the patients had an internal plantar flexion moment about the ankle at foot-contact on the short side (Fig. 2), and this was used to define this compensatory strategy objectively.
The compensatory strategy of persistent flexion of the long limb implied that the average flexion and extension moments about the hip or knee were more than one standard deviation greater on the long side than on the short side (Fig. 2-B). Children who were visually classified as persistently flexing the long limb did indeed have greater peak flexion of the hip (p < 0.001) and knee (p < 0.001) and greater extension moments about the hip (p < 0.02) and knee (p < 0.02) on the long side. They had a smaller peak flexion moment about the hip (p < 0.03) and greater flexion of the knee (p < 0.007) and hip (p < 0.001) on the long side than did children who did not use this strategy. Four children who were thought initially to use this strategy on the basis of the videotape did not meet the objective criteria and were reclassified: two used circumduction, one used vaulting, and one did not use any detectable strategy.
There was no difference in flexion or extension of the hip or knee, extension moments about the hip or knee, or vertical displacement of the center of mass in the children who were visually classified as using vaulting compared with those who did not use this strategy. The children who were visually identified as using circumduction did not differ from the other children with regard to abduction, adduction, or rotation of the hip; abduction and adduction moments about the hip; or hip powers in the coronal plane.
The range of measurable limb-length discrepancy was 0.6 to 11.1 centimeters, or 0.8 to 15.8 per cent of the length of the long extremity (Fig. 1). The per cent difference in limb length between the short and long extremities did not vary among the four methods used to measure limb length. The relative contribution of the femur and tibia to the inequality varied and was not found to be a factor in our analysis. The average discrepancy for the seven patients who had no observable compensatory strategy was 1.64 ± 2.83 centimeters (2.2 ± 4.5 per cent). The average discrepancy for the children who used toe-walking as a compensatory strategy was 6.54 ± 2.83 centimeters (10.4 ± 4.5 per cent) (Fig. 3). The threshold discrepancy associated with the use of toe-walking was 5.5 per cent (p = 0.05).
Eight children had persistent pelvic obliquity (average, 7.8 ± 1.4 degrees), with dipping of the pelvis on the side of the short extremity, throughout the gait cycle. Nine children had a range of pelvic obliquity of more than ±2.37 degrees, which is one standard deviation above the pelvic obliquity seen in normal adult controls tested in our laboratory. Two children had both an increased range of pelvic obliquity and persistent pelvic obliquity, which were not correlated with the degree of limb-length discrepancy (Fig. 4).
The average walking velocity was 1.23 ± 0.16 meters per second. The differences in velocity among the children were minimum, allowing for comparison of the kinetic data among patients. The differences between the long and short extremities with regard to most dependent kinematic, kinetic, and cadence variables appeared to be significant. However, no Spearman correlation coefficient between the absolute or per cent limb-length discrepancy and the difference between the long and short extremities with regard to any of the dependent variables was more than 0.8. Thus, there was no predictive relationship between the magnitude of the discrepancy and the performance of the two extremities. In the group of seven children who had a limb-length discrepancy of 6 per cent or greater, the Pearson correlation coefficient was more than 0.8 (p < 0.05) for greater generation of total hip power and decreased negative work (energy absorption) on the long side. Over-all, the long limb performed more mechanical work than the short limb (Table 1), and the differences were significant at the hip (p < 0.01), knee (p < 0.02), and ankle (p < 0.003) and for the entire limb (p < 0.0001). When we examined the differences in work according to the compensatory strategy used, we found that patients who walked plantigrade, used circumduction, or vaulted over the long extremity had no difference between the extremities with regard to the amount of mechanical work performed. Patients who used toe-walking had more asymmetry and performed more total work with the long limb than with the short limb compared with patients who walked plantigrade (p < 0.0001), with the primary differences at the hip (p < 0.02) and knee (p < 0.001) being significant. Patients who walked with persistent flexion of the long limb had more total work done by the long limb (p < 0.01) as well as more work at the hip (p < 0.008) and knee (p < 0.01) than those who walked with full extension of the long limb. These differences were significant for the seven children who used both persistent flexion and toe-walking (p < 0.01). We could detect no significant difference between the long and short limbs with regard to the mechanical work performed in the twelve children who used persistent flexion but not toe-walking.
Testing in our laboratory demonstrated that one standard deviation for normal displacement of the center of body mass in adults is +1.9 to -2.1 per cent of body height. In the present study, five of the nine patients in whom the displacement exceeded this range used toe-walking as a compensatory strategy. The average vertical displacement of the center of body mass was 2.9 ± 0.55 per cent of body height for patients who used toe-walking and 2.1 ± 0.57 per cent of body height for patients who walked plantigrade; this difference was significant (p < 0.0001). The threshold for increased displacement of the center of body mass was a 5.5 per cent limb-length discrepancy (p = 0.05). We found that the medial-lateral acceleration of the center of body mass was greater for the short extremity at the point of foot-contact, which suggested more rapid transfer of mass to the short limb. However, this was not reflected by abnormal shear forces at the foot or by different rates of loading between the extremities.
Limb-length discrepancy is a common clinical finding. Rush and Steiner32 found that as many as 70 per cent of 1000 consecutive non-selected adult men had some degree of discrepancy. Edinger and Biedermann7 found that 45 per cent of 325 subjects had a discrepancy of more than five millimeters. Hult20 reported that 30 per cent of 1137 Swedish laborers had a discrepancy of 1.0 to 1.5 centimeters, 4 per cent had a discrepancy of 2.0 to 2.5 centimeters, and 0.7 per cent had a discrepancy of as much as 4.5 centimeters. Some authors have claimed that a limb-length discrepancy leads to mechanical and functional changes in gait27,35 and increased energy expenditure3. Treatment has been recommended for discrepancies of less than one to more than five centimeters3,17,26,27,35, but the rationale for these recommendations has not been well defined. The degree of limb-length inequality that may cause functional problems and the mechanism by which it affects gait and stance have been the subject of few studies. Gross found no noticeable functional or cosmetic problems in a study of seventy-four adults who had less than two centimeters of discrepancy17 and thirty-five marathon runners who had as much as 2.5 centimeters of discrepancy18. Kaufman et al.22 reported asymmetrical gait, as measured by a force-plate, at a threshold discrepancy of 3.7 per cent in a study of twenty children who had a limb-length discrepancy. Greater heel-contact forces for the long extremity33 and correctable supination deformity of the foot of the short extremity5 have been demonstrated in studies of adults who had a discrepancy of two centimeters or less. Small discrepancies have been shown to alter joint moments and powers36 as well as postural sway25. The seven patients in the present study who did not appear to use any compensatory strategy had an average limb-length discrepancy of 2.2 ± 4.5 per cent and did not have any demonstrable kinematic or kinetic asymmetry of gait. We believe that treatment of these small discrepancies to prevent asymmetrical gait is not warranted. Morscher26 suggested that limb-length discrepancy resulted in persistent pelvic obliquity during double-limb stance and that larger discrepancies led to greater obliquity. He believed that pelvic obliquity led to uncovering of the femoral head of the long extremity with an abnormal concentration of forces at the acetabular margin accelerating degenerative osteoarthrosis of the hip. Similarly, pelvic obliquity with lateral sacral inclination has been thought to require additional work by the abductor muscles of the hip21 and the lumbar paraspinous muscles, placing extra strain on spinal ligaments and leading to fixed spinal deformities10-13. Phelps et al.31 reported persistent obliquity of 10 to 20 degrees in seven patients who had a limb-length discrepancy of two to six centimeters, but they did not provide information about associated angular, rotational, or neurological abnormalities. We did not find a correlation between limb-length discrepancy and pelvic obliquity or abductor muscle strength, as measured with Cybex testing, in our patients. Only eight of our thirty-five patients had persistent pelvic obliquity during gait, and the average pelvic obliquity during gait was no different than that of normal controls. Although pelvic obliquity can occur in association with a limb-length discrepancy, we believe that it is not a common finding and that most otherwise healthy individuals develop compensatory strategies to maintain a level pelvis during gait. The compensatory strategies observed in the present study were equinus positioning of the ankle of the short limb (toe-walking), vaulting over the long limb, increased flexion of the long limb, and circumduction of the long limb. These strategies have been suggested by other authors3,22,27,31; however, we were unable to find any objective description of the movements. We were able to define objective criteria for the strategies of toe-walking and increased flexion of the long limb. We initially believed that circumduction would be evident as increased abduction and external rotation of the hip during the swing phase of gait and that vaulting would result in an increased vertical translation of the center of gravity of the body between the single-limb and double-limb phases of stance for the long and short extremities. However, we were unable to demonstrate these findings in the patients who were thought to use circumduction or vaulting. The lack of definable objective criteria and the poor interobserver agreement regarding the appearance of these two strategies suggest that they are complex movements that are often used in combination with other strategies. The combination of several strategies may shift the timing of kinematic alterations to portions of the gait cycle during which they are not as apparent. Inman et al.21 showed that compensatory strategies dampen oscillations of the center of body mass and decrease over-all energy expenditure during gait. Our patients who used toe-walking had increased mechanical work performed by the long limb and had greater vertical translation of the center of body mass during gait than did normal controls. Total mechanical work is a measure of concentric and eccentric muscle contractions crossing a joint. It does not take into account the energy expenditure of isometric muscle forces, but it is associated with the energy needed to produce the work30. The true metabolic cost of the compensatory strategies used by our patients is unknown since we did not measure this parameter. Phelps et al.31 did not find that patients who had as much as six centimeters of limb-length discrepancy had more oxygen consumption or oxygen cost than normal adult controls. We believe that individuals who have a smaller discrepancy are able to normalize mechanical work and the metabolic cost of walking between the extremities with use of the compensatory strategies previously described. However, when a discrepancy of 5.5 per cent is reached, toe-walking as a compensatory strategy cannot equalize the work performed by the two extremities. We found that most of the difference in mechanical work in our patients occurred at the knee and hip in the sagittal plane. Eng and Winter8 reported that, during gait, 74 per cent of the work at the hip, 85 per cent of the work at the knee, and 95 per cent of the work at the ankle is performed in the sagittal plane. Although pain over the greater trochanter secondary to overactivity of the abductor muscles has been reported in individuals who have a limb-length discrepancy26, we did not find this symptom in our patients. Moseley27 believed that adults do not use the adaptive strategies that are employed by children because the strength-to-weight ratio is greater in children. Nineteen of our thirty-five patients used increased flexion of the long extremity as a compensatory mechanism. We cannot say whether the compensatory strategies used by these children will be maintained in adult life. The lack of differences in work between the long and short extremities in the patients who had a smaller limb-length discrepancy suggests that compensatory strategies provide an efficient gait pattern that may serve these children well as adults. The results of this study are not applicable to adults who have an acute acquired limb-length discrepancy.
NOTE: The authors thank Sameer Kolangaradath and Cindy Smith for their invaluable research assistance and Richard Browne, Ph.D., for his editorial assistance.
*No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study.
  CiteULike  Connotea  Del.icio.us  Technorati What's this?
This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Top
Abstract
Introduction
| home | CME | help | feedback | subscriptions | archive | site index |
|
Stanford University Libraries' HighWire Press (TM) assists in the publication of JBJS Online. |