Development of a scientific approach to the study of socket discomfort could lead to improvements in outcomes because socket discomfort problems are one of the major reasons amputees reject their prostheses and change prosthetists. Socket replacements are also costly. Substantial pressure variations can occur inside an ischial containment transfemoral socket during gait. The research questions addressed here are: How are socket pressures perceived by an amputee on different regions of the transfemoral residual limb? and What is the nature of the relationship between perceived pressures and measured pressures? These questions have relevance to the design and fitting of sockets because verbal reports of pressure and discomfort from the patient are among the primary, initial means by which the prosthetist determines that fit adjustments may be needed. Answers were developed by applying psychophysical methods to a case study of an individual with a transfemoral amputation.
Previous research examined the use of psychophysical methods and signal detection theory (SDT) to quantify pressure-related socket discomfort but without the benefit of having measurements of actual intrasocket pressures.1,2 The rationale for exploring the use of psychophysical methods and SDT is that they provide a theoretical basis and quantitative data for research on socket design, as well as offer the potential for increasing the scientific basis for socket-fit problem solving in a clinical setting. This report discusses collection and analysis of data concerning the subjective magnitude of the pressures experienced by a subject with a transfemoral amputation and examines their relationship to measured intrasocket pressures obtained using a commercially available pressure sensing system.
The previous research used a Borg scale3 to examine perceived pressures on transtibial residual limbs. The Borg scale is a category ratio scale that uses descriptors to anchor the end points. Figure 1 depicts the Borg scale used in this research. Data on subjective phenomena obtained using a ratio scale can, in theory, be analyzed using quantitative methods.4 Pressures for the previous research1,2 were developed in two ways: by enclosing the limb in a pressure bladder, producing a uniform global pressure, and by inserting thin wafers into the socket over bony areas, producing a clamping pressure. Results suggested that for global pressure, as pressure was increased, experienced prosthesis users perceived relatively constant subjective pressures over a range of applied pressures. However, the psychophysical functions relating the magnitudes of the applied global pressure to perceived pressure, and perceived pressure to discomfort differed among the individuals and were unique for each person. Residual limb morphology, amputation surgery, and experience were hypothesized as explanations. It was found that when subjective pressures were reported to be around 3 on the Borg scale (moderate), discomfort began to be experienced. When wafers were inserted into the sockets over bony prominences, discomfort was reported for subjective pressures between 2 (weak) and 3 (moderate) in magnitude, and a strong linear correlation was found between perceived pressure and perceived discomfort.
Among individuals having normal sensations, pressure-induced disturbance of circulation leads to discomfort, followed by pain. Research indicates that when external pressures exceed 32 mmHg, the duration that tissues can tolerate a pressure before showing signs of trauma declines.5 But when applying these findings to sockets, care must be taken to distinguish among the ways pressure can be created. In addition, pressures inside a socket vary over the gait cycle and over the different regions of the socket, creating temporal and spatial pressure gradients that could foster blood flow. Concurrent research results (see “Concepts of Pressure in an Ischial Containment Socket: Measurement” this issue) suggested that pressures created by downward movement of the femur could create a quasi-hydrostatic form of pressure, whereas the abduction-adduction and flexion-extension moments necessary for pelvic and anterior-posterior (A-P) stability could create more of a clamping type of pressure, particularly where tissues are thin and lie close to the skeleton. Based on the principles of engineering mechanics, these pressures would add together.
The most frequently cited experimental studies of the effects of pressure on blood flow have examined loading of a clamping nature that involves compressing layers of tissue between a bone and a hard surface or pressing a plunger into tissue.6,7 Under a clamping load, vessels are not deformed uniformly around their circumference but decrease in diameter in a direction parallel to the force, which can reduce the passage of cellular material in the capillaries and alter perfusion. Shear force must also be taken into consideration because the presence of shear reduces the magnitude of normal pressure required to occlude capillaries.8 In a transfemoral socket, the brim, femoral relief and lateral wall, ramus, adductor longus tendon, Scarpa’s triangle, gluteal fold, and ischial regions can generate clamping forces because of relative thinness of the tissues overlying the skeleton and tendons. Bennett9–17 has published a number of theoretical and experimental articles on the effects of linear brim-type loading on stress distribution in tissues.
In regions of highly mobile adipose tissue where there are thicker layers of tissue between the socket wall and the underlying skeletal structure, such as the distal muscle compartments of a snug transfemoral socket worn by an individual with abundant distal adipose tissue (i.e., the subject in this study), a different type of stress distribution may exist.18 In these regions, pressures generated near the socket wall because of downward movement of the femur deep inside the limb may act more uniformly on the circumferences of the vessels in a quasi-hydrostatic manner. Forces originating at the femur would be distributed outward toward the wall of the socket. If the forces were absorbed by thick layers of adipose tissue with low shear resistance, it is possible that the stress distribution could lead to tissue displacements that would tend to distribute forces more evenly and uniformly. If this were the case, it is possible that the surrounding matrix of tissue, assumed to be incompressible, could even absorb a high uniform stress without collapsing the capillaries, much like tunnels can be bored deep in the earth or holes can survive in gels that are loaded uniformly. While this model is conjectural in the context of socket biomechanics, there is evidence in support of it in other contexts.19 If this model is correct, then under a uniform pressure distribution, the distortion of vessels and capillaries might be less, and higher pressures could be tolerated for longer periods. In such cases, traumatizing stresses would be most likely to occur where external pressure gradients produced by socket geometry change steepness rapidly over short distances and create tourniquet effects on blood vessels. Excess or insufficient reduction of socket circumferences during rectification, as well inflexible brims or brims having nonoptimal radii, could produce rapid changes in pressure gradients and high internal stresses in tissues.
The mechanism by which the pressure-sensing mechanoreceptors respond to loading in clamping and uniform pressure situations also is different, which could result in different types of neurophysiologic response and different sensations of pressure magnitude. Pacinian and Ruffinian mechanoreceptors are sensitive to clamping pressure. Finite element studies suggest that when a nerve fiber is compressed by clamping forces, the cross-section of the endoneurium covering the fiber will change shape from circular to elliptical, and it will stretch.20 The resulting increase in its circumference is likely to affect its permeability and electrical properties and could trigger firing, which would be experienced as pain. In contrast, nerve fibers subjected to uniform circumferential pressure elongate but remain circular. The maximum displacement occurs at the edge of the area where pressure has been applied. The longitudinal shear created also reaches a maximum at this edge and, together with the rapid change in pressure gradient at the edge, can cause occlusion of vessels supplying the nerve. The magnitude of discomfort perceived would be related to the amount and duration of blood flow occlusion, which is a function of the magnitude and duration of the pressure. Thus, a uniform circumferential pressure on a mechanoreceptor might not be perceived by an individual as having the same subjective magnitude as an equal amount of clamping pressure as long as there were no major, abrupt changes in the pressure gradient along the axon and no underlying pathology that sensitized the nerve. This also is conjectural, but if it were the case, an individual might not be able to easily distinguish pressure differences that exist between the muscle compartments at a given point during gait, unless the magnitude of the difference is large. However, during walking the force exerted by the femur to maintain A-P and medial-lateral (M-L) stability would be superimposed on the force created by the downward movement of the femur and would act directionally to press the tissues against the socket wall, much like a clamping force. These forces also would vary over the gait cycle. Thus, during gait with a snug socket, there would be a superpositioning of clamping pressures created by the femur and skeletal structure and more uniform pressures created by downward movement of the femur acting as a piston to displace the more distal tissues.
SDT posits that the ability of an individual to detect a neural signal indicative of a change in the state of an external stimulus, such as a difference in pressure (e.g., a perception that pressures in adjacent regions of the limb are different or they are changing over short periods of time), is a function of the neural capability of the brain to discriminate among incoming neural messages together with a criterion set by the individual on the basis of the perceived benefit and cost of detecting a difference or change.21 According to SDT, the magnitude of the difference or change in pressure and the expectation that a difference or change will occur also influence detection.
During gait, stability is of paramount importance. Because the socket is the coupling between the amputee and the external ground reaction force produced at the bottom of the prosthetic foot, it is likely that the detection and interpretation of spatial or temporal differences in socket pressure play a role in maintaining stability. The information processing capabilities of the brain (number of channels) are limited, and it requires time to process the information transmitted to it.22 Generally, larger amounts of information require longer processing times. During gait using a prosthesis, potentially large amounts of pressure information may be transmitted from the socket to the brain. When a message-rich environment exists, the brain sets priorities on what information to process and bring into the stream of consciousness and what information to ignore. In nonamputee gait, much of the priority-setting occurs below the level of consciousness, and much information processing occurs in the spinal cord. In the case of amputee gait, with information on stability now coming from pressure sensations in the socket, rather than the proprioceptive information normally transmitted from the mechanoreceptors of the anatomical feet and joints, the brain may be forced to reallocate information processing resources among incoming neural signals. The choice may concern detection and interpretation of spatial differences in pressure (differences that exist between different regions of a socket at a single point in time) versus incoming neural signals that detect temporal changes (differences that correlate to phase of gait). Because of the lack of previous research on this subject, the trade-off in importance to an amputee of the need to discriminate between temporal versus spatial socket pressure differences is unknown. It should be pointed out again that this is conjectural, but it is not unreasonable to hypothesize that the rapidly changing dynamic forces and pressures during gait and the importance of maintaining stability may bias the allocation of information processing resources toward the detection and evaluation of temporal changes. However, if discomfort or pain were to occur, an individual might shift some portion of attentional resources toward the location of the pain and the processing of information concerning spatial differences.
A subject was recruited who had sustained, as a result of trauma, a transfemoral amputation on the left limb just proximal to the femoral condyles. The subject had been fitted with an ischial containment socket featuring a flexible inner socket and frame with windows, and the prosthesis featured a hydraulic single-axis knee, an energy storing foot, and a torque absorber. The subject had worn the socket for 6 years and was pleased with the fit, which was described as being snug, but comfortable. However, the subject did experience trauma occasionally at the cut end of the femur on the lateral side because of a very thin covering of tissue as a result of the surgical technique. The subject’s femur was 22 cm long and featured a moderate amount of unanchored loose tissue that offered considerable longitudinal and rotational mobility with respect to the femur. It could be characterized as offering little resistance to shear, except at the cut end of the femur on the lateral side. Evaluation of the subject’s sensitivity to touch was performed using a biothesiometer, and no impairment was detected. The subject was very active, and went swimming, scuba diving, cycling, and hiking using a prosthesis. The preferred donning method used by the subject was to pull into it. No problems with loss of suction were reported, and the alignment of the prosthesis and gait of the subject appeared very good. The subject read and signed a letter of informed consent that had been approved, along with the human subject protocol, by the Biomedical Sciences Committee of the Institutional Review Board of the University of Nevada, Las Vegas.
Pressure measurements were taken at the residual limb-socket interface during normal gait using F-Socket (Tekscan, Inc., South Boston, MA). The socket was disaggregated into regions representing the muscle compartments and smaller areas of high pressure. The data obtained from the F-Socket sensors subsequently were windowed to represent the compartments, and mean average pressures were computed over five consecutive steps for each windowed region.
During collection of the pressure data, the Borg scale shown in Figure 1 was shown to the subject, who was asked to use the scale to report the magnitude of the pressure being experienced for each region of the socket. Questions were asked by the researcher to ascertain as precisely as possible the phase of gait during which the maximum and minimum perceived pressures occurred, as well as the subjective magnitude. In some cases, the subject reported the sensation of the maximum or minimum pressure seeming to occur not at a precise instant, but over a brief span of time associated with the beginning or ending of a phase, and ending or beginning close in time to another phase. In these cases, subjective magnitudes were recorded for both ends of the phase, which meant that similar numbers were recorded for two points in time. The subject, who was trained in a health care specialty, expressed no difficulty in either providing subjective scale values or understanding the locations of the socket and anatomical features for which evaluations were sought. However, the task did require concentration, and the subject usually walked several additional steps while focusing on the pressure sensations.
Subjective responses are presented in Table 1. Although the subject reported no discomfort for any region of the socket except the lateral wall in the region of the femoral relief, results indicated that the subject did not perceive equal pressures in all the muscle compartments. Generally, the same perceived magnitudes were perceived for the proximal and distal compartments during each phase of gait, but different values were reported for the posterior and anterior compartments. Among the distal compartments, the subject correctly identified the phase of gait during which a pressure maximum or minimum occurred in all but the antero-lateral compartment, for a correct detection rate of five of six (83%). In the proximal muscle compartments, the correct phase of gait for pressure maximums and minimums was detected in four of six compartments, for a match rate of 67%. However, for a fifth compartment (proximal anterior), the phase associated with maximum pressure was correctly identified, whereas the phase of minimum pressure was incorrectly identified. For both proximal and distal antero-lateral compartments, the phases of gait were incorrectly identified.
In the regions of expected high pressure (Scarpa’s triangle, femoral relief, lateral and posterior gluteal fold, ischium, ramus), the identification of the phases of gait associated with both maximum and minimum perceived pressures correctly matched the phases of gait associated with both maximum and minimum mean average pressures at only one location, the ramus. Phases associated only with peak pressures matched subjective responses at two additional locations, the femoral relief and Scarpa’s triangle. In the gluteal fold and ischial seat region, reported pressures varied inversely with measured pressures; terminal stance produced the highest recorded pressure measurements, but the subjective responses were the lowest. Thus, the overall correct detection rate for the regions of high pressure, one of seven or 14.3%, was not as good as for the muscle compartments, with the posterior region of the socket showing the greatest discrepancy between measured and subjective values. Taken together, the femoral relief and the ramus had the greatest number of correct detections. At the femoral relief during the maximum measured pressure, which occurred at midstance, a subjective pressure of 3 (moderate) was reported. This was the highest subjective magnitude of pressure reported anywhere in the socket, and it was the region that the subject indicated produced discomfort. At the ramus, although average measured pressures were roughly double those of the femoral relief, a subjective pressure of only 1 was reported.
As stated, pressure differences also existed between the muscle compartments of the socket during each phase of gait. One of the questions examined was whether these differences were detected correctly at a higher or lower rate than pressure differences occurring during gait (spatial versus temporal pressure difference detection). Table 2 contains responses concerning correctly perceived differences in pressure between adjacent muscle compartments during the loading response phase of gait (when pressures increase rapidly inside the socket) and correctly perceived differences in pressure between the phases of gait when maximum and minimum pressures occur. The frequencies indicate that the magnitudes of the pressure differences tended to be greater between adjacent muscle compartments during loading response (spatial) than between phases of gait within muscle compartments (temporal). However, the proportion of correct detections was greater for the temporal differences than for the spatial differences.
Ideally, psychophysical functions that relate actual pressures to perceived pressures would be developed using statistical methods and subjective data of the type obtained. However, the experimental design under which the data were collected precluded this. To statistically construct psychophysical functions, the procedure would involve experimentally varying the pressures at a given location in the socket over a range of values by means of socket modifications and taking measurements over multiple trials. This was not feasible given the amount of time the subject had available for research. Instead, a plot of pressure versus subjective magnitude was constructed for each region of the socket to determine the possible shapes that the underlying psychophysical functions might take.
Figure 2 is a plot of mean pressure versus subjective magnitude for the points where responses were in agreement with the maximum and minimum measured pressures. Each region is represented by a different line or a single point because it was hypothesized that the psychophysical functions could vary from region to region on the limb. These points represent socket regions and phases where correct detection occurred and show how the subjective magnitude of perceived pressure varied with measured pressure when detection was correct. The remaining points not shown constitute either detection error or reflect the lack of precision with which the subject was able to identify exactly when maximum and minimum pressures occurred. Although this is useful information in the context of SDT and can be used to develop a receiver operating characteristic curve indicating the correct detection rate as a function of the criterion set by the subject, the experimental design precluded this more extensive analysis.
The subject was an experienced prosthesis user, had worn the same socket for many years, and reported it as snug but comfortable, with the exception of the distal lateral femoral region. The subject reported very low subjective magnitudes of pressure in the muscle compartments, even though measured pressures were of a magnitude that might cause concern about circulation. If circulation were impaired, it is likely that discomfort and pain would have been experienced.
A question arises as to whether the pressure gradients in a transfemoral socket might influence blood flow. A potential source of pressure is the femur, which in a snug socket will generate internal forces if it moves downward during loading response and directs forces posteriorly and laterally as the hip extends and abducts. This could create pressure gradients during gait, which might vary dynamically as pressures vary. Thus, socket design might have an influence on pressure gradients related to venous return.
If interface pressures in the muscle compartments decrease in the direction from distal to proximal, this could facilitate the return flow of blood and lymph toward the heart. If the internal pressure gradients increase from distal to proximal, muscle activity or other mechanisms might be needed to overcome this to move fluids back toward the heart. The measured pressure gradients at the limb-socket interface for the subject in this study decreased from distal to proximal in the antero-lateral, lateral, postero-lateral, and posterior compartments, possible contributing to the sense of comfort. The opposite pressure gradient existed for the anterior compartment, and the subject reported frequently developing a feeling of fatigue in that muscle compartment by the end of a day.
With respect to inflow into the capillaries from the arteries, another hypothesis is that the capillaries in tissues might not collapse under a uniform pressure. If this hypothesis has any validity in the context of the hemodynamics for a residual limb that fits snugly into a socket, it might also help explain why pressures as high as those measured did not produce discomfort in the muscle compartments. However, all of these proposed explanations are conjectural. There is a scarcity of literature on the subject. It is noted that the noninvasive measurement of blood flow in a residual limb enclosed in a socket, by methods such as Doppler ultrasound, photoplethysmography, thermography, or transcutaneous oxygen tension, is difficult because of interference caused by the socket materials.
Concerning the consistency of subjective measurements obtained using the Borg scale, four subjects in two different studies,1,2 representing users of both transtibial and transfemoral prostheses, have associated subjective socket pressures of magnitude 2.5 to 3 (between weak-moderate and moderate) with discomfort problems. This finding suggests, tentatively, that a subjective pressure of a magnitude of 2.5 or higher reported during the fitting of a check socket might be an indication that a possible discomfort problem could develop with the socket at that location. This requires additional research.
The subject did not report the fairly large pressure differences existing between adjacent muscle compartments during loading response (spatial discrimination of pressure differences or contrasts), whereas smaller pressure changes were detected between phases of gait (temporal discrimination of pressure differences or contrasts). This supports but does not prove the hypothesis that the need to process information on temporal changes related to stability may be assigned a higher priority by the brain than information on spatial differences. Over time, as long as circulation is not impaired and there is no discomfort, individuals may adapt to spatial patterns of pressure sensation in a snugly fitting socket and forego development of the ability to discriminate small spatial pressure differences because this is not essential to gait. However, alternative hypotheses might be advanced. One hypothesis is that the ability to discriminate spatial pressure patterns and differences is a function of the density, location, and responsiveness of the mechanoreceptors in the residual limb, and these vary by region. Because the task in this experiment asked the subject to make temporal, rather than spatial, discriminations, the first hypothesis cannot be explored further.
Based on the locations of the points and slopes of the curves, Figure 2 suggests that several different psychophysical functions may be involved for different regions of the limb. The lines for all the muscle compartments except the lateral appear to be grouped together, with differences in slopes. The lines for the proximal compartments generally are steeper than the lines for the distal compartments, suggesting that the proximal compartments may be more sensitive to pressure than the distal ones, which could be related to the proximal and distal distributions of adipose tissue and possibly the manner of loading. The subjective responses for the proximal and distal lateral compartments appear similar, and the point for the femoral relief lies close to an extension of the line for the distal lateral compartment. The subject reported the highest perceived pressures at the femoral relief, and the clamping mode of pressure in this region, as well as in the lateral compartments, could explain these perceptions.
The points representing the posterior gluteal fold and ischial seat during loading response also lie close to these lines. In the posterior gluteal fold region, the subject reported lower pressure sensation during terminal stance than during loading response. Although it seems intuitive that a clamping mode of pressure would be involved during both phases of gait, the magnitudes of the subjective responses during terminal stance are similar to those of the muscle compartments under the hypothesized uniform mode of loading, and during loading response are similar to those of the lateral compartments, where a clamping mode of pressure occurs. One possibility is error in the pressure measurements; however, the pressure data were in agreement with Radcliffe’s model.23 A second possibility is error in signal detection. However, the relationships for both regions, which are close together spatially, are the same, and the differences in reported magnitudes are large compared with other regions of the socket. If error of detection were involved, it could be expected that the subjective magnitudes reported and shown at each end of the lines would be closer in value.
During loading response, the muscles and tissues in the gluteal region are stretched, and there may be less tissue thickness between the socket and underlying bone or tendon than during terminal stance. As gait progresses from loading response to terminal stance, the muscles in this region contract and the tissues increase in bulk as the hip extends. The thickness of tissue between the socket and the underlying stiff tissues (i.e., bone) increases. If there is a simultaneous decrease in the amount of clamping loading on the mechanoreceptors and an increase in the circumferential form of loading, it might be perceived as a decrease in pressure. The relationship for the ischial region appears similar.
These findings may have clinical significance during fitting. If an individual perceives pressure in the gluteal fold and ischial regions to decrease toward the end of stance when pressure actually is increasing, he or she may be less likely to identify the region as a potential problem area of the socket until it has been used for some time. This is a region that patients frequently complain about several weeks after an apparently successful fitting.
The pressures measured at Scarpa’s triangle and the ramus, while nearly double and triple the pressures measured in the distal compartments, did not evoke higher subjective magnitudes than did those in the distal compartments, suggesting underlying psychophysical relationships different from those in the muscle compartments. The underlying anatomy of the Scarpa’s triangle consists of ligament, vessels, and the edges of muscles, meaning that although the pressures would be of a clamping nature, they were not applied directly against a sharp bone, as in the case of the cut end of the femur. At the ramus, tendon and round bone exist. However, at the ramus a comparison may be made to the forces generated by a narrow bicycle seat; although the forces are high, many individuals can tolerate a well-designed bicycle seat for long periods. In a well-designed socket, an experienced user might not find higher pressures objectionable.
Additional research should be conducted on the relationship between mode of pressure loading (clamping versus circumferential) and perceptions of pressure. Gel liners attempt to reduce the magnitude of clamping types of loads via an elastomeric material that exhibits some degree of flow under pressure and thereby helps reduce concentrations of high pressure.24 Additional research also should be conducted on quasi-hydrostatic loading mechanisms and their possible influence on pressure perception. Some prosthetists believe that a well-fitting socket should produce a feeling of uniform pressure. From the perspective of engineering mechanics, a body surrounded by a uniform pressure would have a zero resultant force acting upon it and would be in equilibrium; there would be no resultant moment, and it would neither accelerate nor decelerate. Residual limbs do accelerate and decelerate during the gait cycle, so the resultant force on the residual limb must be greater than zero, which implies that pressure differences must occur. The subject in this study perceived pressure differences related to phase of gait, but this did not prevent the socket from being judged as comfortable. In fact, a research question could be raised on whether pressure sensation during gait might play an important role in providing feedback on stability. It is noted that during any single phase of gait, pressures were perceived to be uniform in the muscle compartments.
The study demonstrated that an assessment of socket fit under static conditions of loading, in the absence of gait, may not give an accurate indication of the range of pressures that will be generated during gait and how the socket will perform or feel when the individual ambulates. Preliminary evidence based on the single subject involved in this study suggests that individuals also may have trouble perceiving pressure magnitudes accurately. Spatially, pressures could vary substantially within a socket during a given phase of gait, but be experienced as having similar subjective magnitudes. Despite this, the ability of the subject to detect pressure changes occurring during gait appeared good for the muscle compartments, which may reflect the importance of these sensations to the maintenance of stability. It is possible that when donning a new and unfamiliar socket, an experienced subject might have a heightened awareness of spatial differences in pressure.
Although the presence of pain during the fitting of a check socket may be a strong indication of excess pressure, the absence of pain during a fitting may not necessarily mean that pressures are in a safe range. It may be that the viscoelastic tissues of the limb have not had adequate time during the fitting session to creep and produce blood flow occlusion for a duration sufficient to cause discomfort or pain.
Research of this type requires a subject who is willing to devote substantial amounts of time to the effort and is motivated to focus on a highly repetitive task that requires accuracy in assessing subjective stimuli. The experiment indicated that the task required substantial concentration by the subject, and the estimation of pressure magnitudes was subject to detection error. To account for the error using standard statistical techniques such as regression analysis would require a much longer experiment that would probably exceed the motivational levels of most subjects. However, the research addressed a very important problem faced daily in clinical practice: how to direct questions concerning the sensations of socket fit to patients and interpret their verbal responses.
A number of interesting hypotheses emerged from the study. A risk exists that some of the data may have been overinterpreted and hypotheses advanced that are incorrect, but it is hoped the questions raised will stimulate others to undertake research that will improve understanding of the biomechanical and psychological variables and processes that underlie the perception of socket discomfort. The influence of socket pressures on circulation, comfort, and stability during gait is an extremely important subject that is worthy of additional research. Little information is available on this subject in the orthotics and prosthetics literature.