Abstract
The paper deals with the identification of motor strategies adopted by trans-femoral amputees to compensate for the constraints of hip motion induced by the interference of the socket with the pelvis and, particularly, with the ischial tuberosity. A group of 11 subjects with trans-femoral amputation, three of whom wore two different prostheses, giving a sample size of 14 cases, were studied by gait-analysis protocols: the present paper focuses on the pelvis – thigh kinematics at foot strike. The results showed that, at the prosthetic side, the hip is significantly less flexed and less extended, respectively, at the ipsilateral and contralateral foot strike. Moreover, the pelvis is significantly more anterior tilted at sound foot strike. The anterior step length showed a decreased sound limb anterior step in 12 out of 14 cases. The authors interpret these results as a combination of mechanical constraints and compensatory actions: the reduced prosthetic hip extension is determined by the mechanical constraint involved in the pelvis – socket interference; and the increased pelvis tilt and sound hip flexion occurring at the same time are compensating strategies, adopted by the amputees, in order to obtain a functional step length and symmetrical thigh inclinations. Those factors determine a gait pattern which is functional, only slightly slower than normal gait, and without any perceivable alterations. On the other hand, the authors show that the increased pelvis tilting necessarily overloads the lumbar tract of the spine and may be related to the frequent occurrence of low-back pain in amputee subjects, despite the positive functional gait recovery.
Introduction
A lower-limb prosthesis for a trans-femoral amputee is generally designed integrating different modules: foot, pylon, knee, and socket. The socket, which encompasses the greatest part of the stump, has two main functions: to rigidly connect the prosthesis to the subject's stump and to allow dynamic exchange between the subject and the prosthesis (Inman et al. 1981). The former function should be fulfilled, avoiding either an excessively tight compression of the stump (Goh et al. 2003), possibly causing discomfort, pain, and local lesions, or a loose interface, which causes the socket to lose contact. The latter function is fulfilled by a properly designed contact surface, particularly in the socket proximal section, between the socket and amputee's body. Generally, in trans-femoral amputees, this contact surface is not limited to the stump – socket interface but includes also a contact between the socket and the pelvic segment: the present paper focuses on the ischial tuberosity as the relevant contact point. This solution is required by the impossibility of having the stump, and particularly its soft tissues, bear the body weight during gait stance phase, but it obviously introduces a relative hip motion constraint which occurs during hip extension. The individual design of the socket is obtained by varying the position of the contacting surface: a lower position tends to unload the ischium and to overload the stump soft tissues, while an upper position tends to constrain hip extension. In other words, the socket – ischium interface plays a double role: the positive role is to bear the most part of the load transfer during the stance phase; the negative role is that, due to its location relative to the hip joint centre in the sagittal plane, the ischium opposes the hip extension.
In scientific literature (see review by Rietman et al. 2002), the analysis of trans-femoral amputee locomotor function has mainly focused on the effect of the prosthetic foot and knee, while the socket has been mainly considered for comfort issues (Neumann 2001a; b).
In the technical field, the innovation in socket design has been mainly devoted to materials, the CAD-CAM process, possibly assisted by objective measurement of the stump geometry (Torres-Moreno et al. 1991).
With regard to biomechanical features, some papers have dealt with the pelvis-stump mechanics for frontal and transverse planes (Gottschalk et al. 1990; Michaud et al. 2000; Sjodahl et al. 2003; Twiste and Rithalia 2003), while other papers have approached the quantification of the motor strategies by adopting joint-related measurements by electrogoniometers (Boonstra et al. 1994; Jaegers et al. 1995) or by static assessment of the trunk – pelvis orientation (Alsancak et al. 1998).
The objective of this paper is to determine the effects on the motor strategies of the body – socket interface. The authors have attempted to explore the hypothesis that the mechanical constraint occurring at the body – socket interface, while being partially managed by modifications of the soft tissue close to the ischium, leads to an adaptation of gait strategy oriented to preserve gait symmetry and, finally, the overall locomotor function. In order to quantify these effects, an experimental protocol, including 3D gait analysis, was performed with a group of 11 trans-femoral amputees wearing prostheses with ischial bearing.
Materials and methods
Eleven subjects (10 male and one female, with ages ranging from 19 to 54 years) were admitted to the study, after being fully informed about its content and giving their consent. All the subjects had a unilateral trans-femoral amputation and use prostheses with ischial bearing. The subjects' gait performance was fully functional to their daily activities. Three subjects were analysed, each using two different prostheses, giving a sample size of 14.
The subjects were analysed using a 3D gait analysis protocol (Frigo et al. 1998) at the Movement Analysis Laboratory at the Bioengineering Centre of the Don Gnocchi Hospital, Milan, Italy.
The subjects performed three walking tests at a self-chosen natural speed and cadence. The measurement was obtained during the steady state of gait. Data analysis produced the full set of variables and parameters for each trial, and finally the individual mean and standard deviation values were considered for the interpretation.
The reference normal data of the listed parameters were obtained from an age-matched control group including seven male subjects, whose ages ranged from 24 to 51 years.
Results
The subjects were able to perform gait trials without any difficulties. Their performances appeared to be qualitatively normal. The amputees' gait speed ranged from 49.0 to 74.1% of body height (BH) per second (the normal reference value is 74.6% BH/s). The stride length was 67.6 – 84.5% of the body height (reference value is 82.0% BH). The cadence, expressed in steps per minute, ranged from 84.5 to 105.3 (the reference value is 109.2). Individual values of cadence and stride length are plotted inFigure 1. Iso-velocity lines are reported as a percentage of normal reference speed.
Scatter plot of reference and cases' individual values of stride length, expressed as percentage of body height, and cadence, expressed in steps per minute. Iso-velocity lines are plotted in correspondence to fixed percentage (100% down to 60%) of reference normal velocity.
A set of gait parameters was computed for both limbs in the sagittal plane: anterior step length, hip sagittal angle and thigh sagittal inclination at ipsilateral and contralateral foot strike, and pelvis tilt at foot strike. In order to demonstrate asymmetries, the differences between the sound and prosthetic limb were computed and are reported inTable I. Those differences are obviously not available in the reference population; nonetheless, normal kinematics can be assumed to be symmetrical, and consequently, the reference value for any difference can be assumed to be null. Moreover, for each parameter, a statistical analysis is reported in Table I, testing the hypothesis that the symmetrical null value belongs or not to the observed distribution of difference values.
Differences of values observed at the foot strike of prosthetic and sound limb in a set of gait parametersa.
P-values are reported for statistical testing of the hypothesis that the symmetrical behaviour (null difference) may belong to the observed distributions of difference values.
In order to show individual absolute values, a scatter plot is presented for each parameter having the prosthetic and the sound limbs, respectively, on the two plot axes. The diagonal line represents the symmetrical behaviour, and consequently, the asymmetry is graphically represented by the distance between the data points and the diagonal line, the larger the distance is, the more asymmetrical the parameters values are. Normal values are reported as segments aligned on the diagonal line. When referring to a parameter that is assessed either at ipsilateral or contralateral foot strikes, the same plot reports both scatter plots, and a dashed line divides the two sectors.
InFigure 2, the prosthesis forward step length appears to be larger than the sound limb one, except in case 8, which had a shorter prosthetic limb step, and case 1, which showed substantial symmetry. The mean difference was not significant as that reported inTable I, but when excluding the two previously cited cases, the statistical analysis demonstrates a significantly longer step for the prosthesis (P=0.04).
Individual forward step length, expressed as a percentage of body height, of the prosthetic (ZP axis) and sound (ZS axis) limbs. The normal reference range of forward step length is plotted as a segment on the diagonal line, marking symmetrical behaviour.
The sound thigh forward inclination is slightly larger, albeit not statistically significant, while the backward inclination at the contralateral foot strike is not significantly different between limbs (Figure 3).
Individual thigh-inclination values, expressed in degrees, of the prosthetic (ZP axis) and sound (ZS axis) limbs. The plot shows both the forward-inclination values at the ipsilateral foot strike (upper-right quadrant) and the backward-inclination values (lower-left quadrant). Normal reference ranges of thigh inclination are plotted as segments on the diagonal line, marking symmetrical behaviour.
Hip angle shows a statistically significant pattern in which the prosthetic limb has a reduced hip motion range: the joint is less flexed at ipsilateral foot contact and is less extended at the contralateral foot contact (Figure 4).
Individual hip-joint angle absolute values, expressed in degrees, of the prosthetic (ZP axis) and sound (ZS axis) limbs. The plot shows both the hip-flexion values at the ipsilateral foot strike (upper right quadrant) and the hip-extension values (lower left quadrant). Normal reference ranges of thigh inclination are plotted as segments on the 45° line, marking symmetrical behaviour.
Pelvic tilt always shows larger values when the sound limb foot is hitting the ground (and the prosthesis is in its push-off phase). Some cases (particularly including the aforementioned cases 1 and 8) are characterized by fairly large values, marking a substantial asymmetry in pelvis motion (Figure 5).
Individual pelvis anterior tilt angle, expressed in degrees, occurring at the foot strikes of the prosthesis (ZP axis) and sound limb (ZS axis). The normal reference range of pelvis tilt is plotted as a segment on the diagonal line, marking symmetrical behaviour.
Discussion
The results presented confirm the impression that gait had an almost normal functionality in the subjects considered; their speed, cadence, and stride length were almost normal, possibly depicting a normal gait slightly slower than the reference ones.
While the global parameters show a functional gait, a perusal of the other parameters leads to identification of asymmetrical patterns which are to be interpreted as a result of the asymmetry of the locomotor systems and the motor strategies adopted by the subjects.
The asymmetry is associated with a set of statistically significant parameters: the pelvic tilt and the joint angles of both hips at the foot strike.
The first asymmetry feature is the reduction in the hip extension by an average of 7.5° on the prosthetic side at the contralateral foot strike. This limitation is directly related to the socket–pelvis interference: the rotation of the femur relative to the pelvis is fixed in the hip-joint acetabulum, while the ischium, and particularly the ischial tuberosity, is not aligned with the previous point in the sagittal plane: this implies that backward rotation of the femur during hip extension is restricted at first by soft tissues and finally by the skeletal constraint.
The second significant feature is the increased pelvis tilt when the hip extends on the prosthetic side. The observed increase of 8.2° is even more significant when considering that the normal range of pelvis tilt during gait is less than 5° (Frigo et al. 1998). Also, the pelvis tilt can be interpreted according to the mechanical effect of the socket pushing against the ischium during hip extension: the pelvis tilt may be adopted to facilitate the backward inclination of the socket or may be induced by the socket pushing against the ischium.
The third significant feature relates to the increased sound hip flexion at ipsilateral contact, while it is worth noticing that the socket did not limit the forward rotation of the femur relatively to the pelvis. It can be shown that increased sound hip flexion is not only due to a larger pelvic tilt. InFigure 6, individual values of increase in pelvic tilt are plotted versus the corresponding values of increase in hip flexion, and the diagonal line is drawn to identify values of increased pelvic tilt that completely determine the corresponding increase in hip flexion. The values plotted show a clear hip-flexion increase, which not only is determined by an increased pelvis tilt but also reveals an additional contribution which appears to be directly proportional to pelvic tilt (correlation coefficient r=0.83). Consequently, the greater sound-side hip flexion appears to be a voluntary strategy.
Increment of pelvis tilt (ZP axis) and hip flexion (ZS axis), expressed in degrees, at the sound limb foot strike compared with the prosthetic foot strike. A strong linear correlation has been identified (r=0.83; P<0.001), and a linear regression line has been plotted as a solid line. The bisecant dotted line is added to demonstrate graphically the relevance of the hip asymmetrical feature compared with that of the pelvis.
The three significant factors identified by a statistical analysis should be classified according to two classes: factors related to functional constraints and factors related to compensation strategies. In this frame, it is obvious that the limitation of prosthetic side hip extension is given by the socket–pelvis interference, a feature which has, at the same time, a positive functional role as load support. On the contrary, pelvic tilt and sound hip flexion are unconstrained actions oriented to compensation.
The previous conclusion quotes the findings reported in a recent paper (Miki et al. 2004) about a different class of patients: an instrumental gait analysis showed that patients suffering from unilateral hip disease showed a reduced hip extension on the affected side and that pelvic tilt increased accordingly: the first feature is said to be related to the hip disease, while the second feature appears to be a compensatory strategy in order to obtain gait symmetry. This interpretation by Miki and coworkers applies also to our analysis of trans-femoral amputee gait.
It is expected that the tuning of this two degrees of freedom compensatory strategy, involving pelvic tilt and hip flexion at sound foot contact, may either undershoot or overshoot in fulfilling the purpose.
The anterior step length is a good outcome index of this tuning strategy: insufficient compensation will result in a shorter sound limb anterior step (reported as a negative value in the authors' approach) while exaggerated compensation will make a longer step: the resulting difference is negative (–4.1°±4.2°) but not statistically significant. This result might be interpreted as a remark that amputees, as a uniform group, are able to apply efficient compensation, but an individual perusal of values may reveal that some subjects are able, and others are not: one subject (case 8) appeared to overcompensate (+5.8°), another subject (case 1) almost perfectly compensated (–0.3°) being characterized by a significantly more dynamic walking strategy (quicker, high cadence, longer strides), thus confirming the finding that speed facilitates symmetry (Miki et al. 2004; Nolan et al. 2003). If those two subjects are excluded from the group, the remaining 12 cases show a significantly reduced sound limb anterior step length.
The two parameters concerning thigh inclination at foot strike reflect the outcome of the compensatory strategy: both indexes do not show any statistically significant asymmetry, thus confirming the impression that amputees are able to produce a gait pattern which is visually perceived as normal. This is particularly true for the thigh backward inclination which is characterized by an almost full symmetry (0.2±3.1°), while the forward inclination shows slightly larger values for the sound limb. A perusal of individual data led the authors to conclude that the two subjects previously identified by their overcompensation (case 8) and by extreme performance (case 1) are those who tend to show the largest asymmetry in this case. It is possible to argue that compensation of expected step length asymmetry in those two subjects is fulfilled by a more forwardly inclined sound thigh, which is a fairly difficult, if not unsafe, action to be performed by a trans-femoral amputee.
In conclusion, the data presented show that the key event in trans-femoral amputee gait is about the beginning of the double support phase with the sound limb forward. When the prosthetic side hip is extended and the ischium–socket interference limits the physiological range, in order to maintain a functional step length the amputees adopt a strategy tuning two independent factors: the pelvis anterior tilt and the sound thigh forward inclination.
Future investigations may explore in more depth the effect of the motor strategies described involving a large pelvic tilt on the lumbar spine kinematics and to correlate such altered motion with low-back pain which trans-femoral amputees often report. It is worthwhile remembering that scientific literature mostly excludes a causal relation between excessive lumbar lordosis and low-back pain (Hansson et al. 1985), but those findings consider static lordosis as measured from static radiographs, while the dynamic lordosis occurring during gait was never considered, and the authors think that a special focus should be applied on this aspect.
The relevant effect of the ischial bearing of the socket in constraining and limiting the hip extension may suggest that the longitudinal position of this socket feature is a critical point: this may represent an issue for future innovation in socket design, possibly introducing some adaptive feature of this parameter, thus defining a ‘smart’ socket.
Finally, the conclusions of the present paper should be verified also with other types of socket for trans-femoral amputees: in particular, the ischial containment design of the socket for trans-femoral amputees avoids concentrating the load on the ischial tuberosity and spreads it across a larger surface, thus diminishing pressure on soft tissues. This reduction in pressure has been demonstrated objectively (Lee et al. 1997) and subjectively in terms of comfort (Hachisuka et al. 1999), but no quantitative descriptions of the modification of the motor strategy compared with other types of socket have been provided.