Strategies for the assessment of pediatric gait in the clinical setting.

SA Rose, BS, PT, is Senior Physical Therapist, Rehabilitation Services and Gait Analysis Newington Children's Hospital, 181 E Cedar St, Newington, CT 6111 USA). Address all correspondence to Ms Rose,

S Ounpuu, MS, is Kinesiologist, Gait Analysis Laboratory, Newington Children's Hospital, and Instructor, Department of Orthopaedic Surgery, School of Medicne, University of Connecticut Health Center, 10 Talcott Notch Rd, Farmington, CT 6032,

PA DeLuca, MD, is Co-director, Gait Analysis Labortory, Director, Cerebral Palsy Service, and Director, Hip and Foot Service, Newington Children's Hospital; Assistant Professor of Orthopedic Surgery, School of Medicine, University of Connecticut Health Center; and Assistant Clinical Professor, Department of Orthopaedic Surgery and Rehabilitation, Yale New Haven Hospital, 333 Cedar St, New Haven, CT 06510.

The value of objective gait analysis in clinical decision making has been demonstrated through applications for surgical, orthotic, and therapeutic treatment planning.(1-10) Normal gait is complex, but pathological gait is even more complex. For example, children with cerebral palsy (CP) have complex gait deviations occurring in three planes of motion, primary deviations are often confounded with compensations, and findings can be missed by observational gait analysis.(11) The gait of children with CP may best be analyzed using a three-dimensional (3D) computerized motion measurement system that can aid the clinician in delineating primary problems and secondary compensations. Coupled with information from other measures of gait assessment, the clinician is then able to make more scientific decisions regarding surgical treatment and objectively evaluate the outcome.(3)

The purpose of this article is to review the various components of computerized gait analysis with an emphasis on the type of information that can, and cannot, be derived from each component. A review of normal kinematics and kinetics with references for phasic electromyographic (EMG) activity and time-distance variables for the pediatric population is provided. Two case examples of children with CP are provided to demonstrate the value of objective computerized gait assessment in neuromuscular disorders, one in which gait analysis information was applied to surgical treatment planning and outcome and one demonstrating the evaluation of orthoses. Finally, guidelines for referral of patients to a gait analysis laboratory will be reviewed.

Gait Analysis Measures,. the Pros and Cons of individual Components

Gait analysis is a broad term that can refer to many different methods of evaluating a subject's walking pattern. The most comprehensive gait analysis today is one that provides an assessment of videotaped recordings, clinical measures, EMG activity, 3D joint kinematics and kinetics, and energy expenditure. Each mode of evaluation provides important information, yet none, we believe, is complete in and of itself. Each component has limitations that should be recognized and understood. Proper interpretation of gait analysis information is possible only with a knowledge of how the information is collected, processed, and calculated. For example, on clinical evaluation, if one were to report the degree of tibial torsion, it would be important to know how it was measured. Measurement by bisecting the malleolar axes may produce different results than evaluating the foot/ thigh angle in the prone position with the knee flexed to 90 degrees. Likewise, the angle definitions on the kinematic plots are based on the marker set used and how it is aligned. in both circumstances, it is critical to understand exactly what information a particular measurement provides in order to make appropriate recommendations based on the information.

Observational and Video Gait Analysis

Observational gait analysis appears to be the technique most often used in clinical settings.(12) Qualitative descriptions of ambulation can be made by observing stability and balance, velocity and control, symmetry and movements of the upper and lower extremities and trunk, weight transfer, foot placement, deformities such as hind-foot varus, and the influence of assistive devices. Such descriptions provide the clinician with general information about the child's overall walking pattern. It is only through observation that we can assess the "functional" ambulation of a child. Function, during gait, may be defined as the ability to ambulate during activities of daily living. Factors influencing functional ambulation include uneven terrain, curbs and hills, and obstacles in the school and home settings. One of the assumptions made with computerized analysis is that the type of walking pattern demonstrated in a laboratory setting is typical of the patients' community ambulation.

This is an assumption that may have the greatest impact on the more severely impaired patient who is able to ambulate over smooth, level surfaces, yet is greatly challenged by the least of environmental constraints. We believe these differences in functional gait can be noted by observational gait analysis.

To aid the clinician in describing gait, documentation forms have been designed to help organize the clinician's observations and facilitate a more comprehensive evaluation of gait.(13,14) After practice, these forms are relatively easy to use, and they can be used across different environments. Many aspects of gait cannot be observed at normal speed. Video analysis is an excellent tool that can be used to supplement observational gait analysis. This analysis allows the clinician more time to observe gait by reviewing the information repeatedly without the effect of patient fatigue. When used in combination with a preset documentation process, video analysis can be helpful in recognizing deviations. A video machine with stop-framing and slow-motion capabilities allows multiple joint-level comparisons during a particular portion of the gait cycle. Sagittal- and coronal-plane motion can be viewed simultaneously when the output of two video cameras are displayed on one screen. One major advantage of observational and video analyses is that encumbering measurement devices or invasive electrodes are not required. These analyses, therefore, may provide the best representation of a patient's typical walking pattern. There is some evidence that equipment may affect gait patterns. Young et al]5 found that temporal and stride variables changed as a result of surface electrode application and changed even further when fine-wire electrodes were used.

It has been shown that two-dimensional (2D) cinematography and the measurement of joint angles using a goniometer on a video screen yield similar results when evaluating sagittal-plane motion of the hip, knee, and ankle in mid-stance in patients with and without pathology.(16) The meaningfulness of a 2D analysis of a 3D activity, however, must be considered. Specifically, neither video nor 2D cinematography account for out-of-plane motion (ie, the combination of eversion that occurs with ankle dorsiflexion during the swing phase of gait). As a result, a sagittal-plane view alone will give misleading results if there are significant coronal- or transverse-plane motions. The error associated with out-of-plane movement in normal gait has been demonstrated and is greatly magnified as the degree of out-of-plane movement increases, as may be seen in children with CP.(17) One major limitation of observational or video analysis is the lack of objective measurement of specific gait descriptors, such as joint angles. The error associated with attempting to measure angles directly from a screen that is a 2D representation of a 3D activity is also a major limitation. Interrater reliability of observational gait analysis is limited even in highly trained clinicians using stop-motion video recording.(12)

Time-Distance Variables

Time-distance variables in gait that can be measured include step length, stride length, cadence, cycle time, and walking velocity. Fairly simple techniques can be used to measure these variables, such as using a stopwatch with a marked walkway and powder on the feet. Step length is the distance from initial contact of one extremity to initial contact of the opposite extremity. Cadence is the number of steps taken per unit of time, and walking velocity is the distance traveled per unit of time. These measures provide information about symmetry in the lower extremities by comparing step lengths and stance-phase/ swing-phase ratios, about stability by examining the time spent in the stance phase versus the swing phase, and about function by examining walking velocity in reference to cadence. That is, a high cadence may indicate instability and decreased function, especially if step lengths are less than normal. Although time-distance measures may be good indicators of overall function, they are descriptors of an end product and do not provide information about the component segments that interact to produce these measures. In this way, these measures do not aid the clinician in determining the origin of the pathology requiring treatment. It is also important to note that variables such as stride length, step length, and walking velocity are related to stature.(18) Therefore, conclusions regarding an increase in function over time based on these measures may be a function of an increase in growth and not the treatment itself. in these cases, the contribution of growth must be factored out.

Sutherland et al(18) evaluated the walking patterns of 309 children ranging in age from 1 to 7 years. They concluded that, whereas a mature gait pattern is established by 7 years of age, time-distance variables vary with age and stature. These variables were reported for each age group from 1 to 7 years and are available for comparison with similar patient populations. Caution should be taken, however, when comparing similar-aged subjects with and without pathologies that may affect leg length.

Electromyography 

The EMG is the electrical signal associated with a muscular contraction.(19) Electromyographic analysis can provide information about the timing and intensity of a muscle contraction. Use of EMGs for the purpose of obtaining timing or phasic information is popular. Phasic data allow us to determine whether a certain muscle's EMG activity is normal, out of phase, continuous, or clonic. This application can prove useful in evaluating the cause of a movement abnormality and aid in planning for tendon transfers.(3,4) For example, if an inversion posturing of the foot is noted during the stance or swing phase, the EMG of the anterior and posterior tibialis muscles may provide information on which muscle is contributing to the deformity. The interpretation, however, may be more difficult if EMG data are analyzed alone, without consideration of whether the muscle's activity is abnormal or in response to the demands placed on the joint. For example, if a nondisabled person walks in a crouched position, the activity of the quadriceps femoris muscle will be noted throughout the stance phase (JR Gage, Newington Children's Hospital [NCH]; personal communication; 1990). If a patient with CP walks with a crouched gait and demonstrates continuous activity of the quadriceps femoris muscle in stance, the conclusion may be incorrectly drawn that this activity is an abnormal spastic response, when the contraction is required to prevent falling.(2)

Because EMG amplitude, under some conditions, has been shown to be correlated with force (ie, higher amplitudes are seen with greater force production), the study of EMG activity during gait has also taken an amplitude focus. In this way, activity level can be related to work.(19) Caution is needed with this interpretation, however, because the nature of the relationship (ie, linear or curvilinear) between EMG activity and force has not been well established. Additionally, studies of this relationship have mainly been limited to measurements taken under isometric test conditions.(20,21) Interpretation of the EMG signal in terms of amplitude across different muscles is difficult. The amplitude of an EMG signal is affected by many factors including electrode location with respect to the muscle of interest, interelectrode distance, subcutaneous fat, and skin/electrode gel temperature.(22) Therefore, to compare EMG amplitudes across muscles, within or between subjecls, the data should be normalized to a reference.(23-25)

The clinical utility of EMG data, therefore, may not lie in the assessment of amplitudes, but rather in identifying the phase activity of muscles. Recognition of the limitations should lead to cautious interpretation of EMG data. Nonetheless, when abnormal activity is noted, conclusions regarding the causes of the abnormality are best drawn after coupling the EMG data to the kinematic, kinetic, and clinical data.

In order to interpret EMG data from patients with pathologies, a reference to normal EMG activity is usually made. Normal EMG patterns for adults(26,27) and children(18,28) are well documented. The inherent variability of EMG patterns within and between subjects should always be taken into account when interpreting EMG data.(26) One should also be aware of the changes in EMG patterns that occur with age in children without pathology before interpreting EMG patterns in patient populations.(18,29)

Electromyographic data can be collected by using surface or indwelling electrodes. Detailed reviews on the advantages and disadvantages of each have been published.(22,30) Indwelling electrodes are more definitive than surface electrodes in terms of sampling activity from a particular muscle, and they are necessary for the evaluation of deep muscles, such as the posterior tibialis muscle. indwelling electrodes are less reliable than surface electrodes,(30) and the location of the wire may migrate as a result of tissue contracting along the length of the wire.(22) When isolated movements are not possible, the only way to be sure of appropriate placement in a particular muscle is through electrical stimulation. indwelling electrodes are also invasive and may promote local muscle cramping and ultimately modify the gait pattern.(15), Surface electrodes are noninvasive and less variable than indwelling electrodes and can provide general information about the activity of large surface muscle groups. (19,23,30,31) The major limitation of surface electrodes is that they are not as selective as indwelling electrodes.(32) The use of surface electrodes, therefore, increases the risk of cross talk, or picking up activity from muscles other than the specific muscle of interest. The larger the muscle of interest, the less the likelihood will be that cross talk from other muscles will be recorded.

Kinematics

Kinematics is a term used to describe movement such as angular displacements of joints and angular velocities and accelerations of limb segments. Kinematic data do not give information about the cause of movement.

The analysis of joint angular displacements using computerized motion measurement systems provides measurements of joint motions.

Common to all computerized motion analysis is some type of reference system such as the use of markers that are placed and aligned with respect to specific anatomical landmarks. This reference system is used to estimate the segment orientations, which in turn are used to calculate the joint angles (Fig. 1). These markers are either active markers, emitting light for camera detection, or passive reflective markers illuminated for the camera by a separate light source. One of the major advantages of passive reflective markers is that they do not require direct connection with a power source and therefore tend to be less encumbering to the subject. The marker alignment illustrated in Figure 1 may differ among gait laboratories. Therefore, the direct comparison of joint kinematic data among institutions must be approached cautiously.

This advantage is of particular importance in the assessment of children. Examples of commercially available systems that use active markers are those marketed by Northern Digital

Incorporated* and Selspot Systems Limited.[dagggar] Some examples of commercially available systems that use reflective markers are those marketed by Ariel Performance Analysis System Incorporated,[double-dagger] Motion Analysis Corporation[section], Oxford Metrics Incorporated,~~ and Peak Performance Technologies Incorporated.# The major advantages and disadvantages of these systems have been described by Whittle.(33) Computer automated tracking, which is available to various extents in all of these systems, has significantly decreased the time required for processing the motion data for routine clinical use. Cinematography, using manual digitization of film frame by frame, was the precursor to the more recent optico-electronic motion measurement systems described. Cinematography, however, has never been a clinically viable tool because of the excessive time required for data processing.

Joint kinematic data may be collected in either two or three dimensions. There are very important differences in the two methods with respect to the way in which the joint angles are calculated and the final data output. In brief, 2D motion systems provide joint angles that are a direct measure of the angles created by the marker set placed on the skin. Normally, 2D data are collected with respect to the sagittal plane. The majority of the data on gait until recently have been calculated in two dimensions.(2,9,19) In a 2D analysis, out-of-plane motion is not accounted for and, if significant, man, introduce substantial error in the sagittal-plane measurements.(17) A similar problem occurs when extracting angles off a video image. The limitations of a 2D analysis are of particular concern in patients with CP who have abnormal motion in the coronal and transverse planes. Sagittal-plane data, however, may be collected in three dimensions and thus account for motion in the other two planes.

When motion data are presented in three dimensions, that is, motion of each joint in the coronal, sagittal, and transverse planes is represented, calculation of the joint angles may incorporate some type of joint centering. joint centering requires the use of a mathematical estimation of the location of the actual joint centers based on the markers and other anatomical relationships that may be measured directly from the patient. Essentially, the marker set that is placed on the outside of the body is "translated" mathematically to the inside so that the final joint angles are a better representation of the actual joint motion. Joint angles are determined for every frame of data and plotted as points across the gait cycle. These points are then connected to produce a curve representing the joint motion through the gait cycle.

Many assumptions and estimations are involved in the calculation of joint kinematics. System users should be aware of these assumptions and estimations and interpret the data accordingly. These assumptions will vary depending on the motion system and the modeling used to provide the joint kinematics. There is also measurement error associated with each system that must be recognized. System error itself is usually believed to be small, because the camera resolution is so high.(28) the major source of error stems from the marker set that is placed externally on the body and thus does not directly represent skeletal motion. The use of bone pins is the best wan, to measure skeletal movement during gait; however, bone pins are not practical for clinical use. The error attributable to skin movement under the markers is difficult to quantify, yet can be insignificant if skin movement is minimal. in general, interpretation of joint kinematics to the nearest degree in any motion system is not appropriate.

The value of kinematic analysis of gait in children with pathology lies in the ability to document joint ranges during function. Kinematic data can be compared with values collected from a population of children without pathology as well as with values obtained before and after a prescribed treatment.(18,28) Interpretation of kinematic curves should be approached with knowledge of the following: (1) the angle definitions based on the particular marker set and marker set alignment, (2) whether the measurements and data reduction are 2D or 3D, (3) the measurement error of the system used, and (4) the trial-to-trial variability of the subject being tested. When comparing a subject's data with the normal curve, it is important to make this comparison to the mean, plus or minus one standard deviation from the mean. The variability of the normal curve is then demonstrated, which helps to avoid overinterpretation of deviations from normal.(34)

The joint kinematic data collected on 124 children without pathology between 5 and 16 years of age at the NCH Gait Analysis Ldboratory are presented in Figure 2. A detailed description of the motion measurement system used to collect these data has been published previousiv.(35) A description of how the joint angles are measured is included in the figure legend.

Kinetics

Kinetics, used in the context of gait analysis, refers to those factors that cause or control movement. The joint kinetic patterns, which refer more specifically to joint moments and powers, are more complex to understand than joint kinematic patterns, because they are not as visual" and require more complex procedures to be determined. The complexity increases if the calculations are made as part of a 3D analysis. Some appreciation for the methods of calculation for joint kinetic data is necessary to be able to understand and interpret joint kinetic patterns. The following is a brief description of the method used for calculation of joint kinetics. For details of these computations, the reader is referred to other published work.(19) When comparing the magnitudes of the moments and powers in all three planes of motion, the sagittal-plane moments and powers are the greatest. In the coronal plane, the moments and powers are greatest in magnitude at the hip followed by the knee and the ankle. It is not yet clear whether there is any clinical significance in the transverse-plane kinetics.(19) Because the majority of power for forward progression appears to occur in the sagittal plane, we will focus on the sagittal-plane moments and powers. The sagittal--plane joint kinematic and kinetic patterns calculated on 31 children without gait abnormalities between 5 and 16 years of age at the NCH Gait Analysis Laboratory are presented in Figure 3. It is important to note that the sagittal-plane kinetic data presented in this article were calculated as part of a 3D analysis.

A joint moment represents the body's internal response to an external load. The net joint moment (what is plotted) represents the sum of all internal joint moments in a particular plane of motion at a specific joint. During normal gait, the influence of soft tissues other than muscles on joint rotations is minimal.(19) Therefore, the moments that are described refer primarily to muscular forces that are acting to control segmental rotation (ie, the algebraic sum of agonist, antagonist, and synergistic muscle groups). As a result, the net joint moment indicates which muscle group is dominant without indicating the contribution of the muscle groups on either side of the joint. In pathological gait, the influence of other soft tissues on joint moments may occur more frequently and more dramatically. For example, a hyperextended knee may be stabilized in stance by the posterior capsule and ligaments. in this case, difficulty arises in determining whether the internal flexor moment is being produced by ligamentous structures or muscular contractions of the hamstring and gastrocnemius muscles. Only if EMG activity is absent in these muscles can one conclude that the forces being produced are of ligamentous origin. Otherwise, the contribution of ligamentous and muscular structures individually cannot be readily determined.

There are two basic methods for calculating joint moments. The first is the ground reaction force (GRF) method. in this method, the resultant GRF is multiplied by its perpendicular distance from the axis of rotation. This method is simple and easy to apply, yet has numerous limitations.(19) With the GRF method, the error associated with the moment calculations greatly increases at the more proximal joints (ie, the hip). The calculations do not incorporate the influence of gravitational and inertial forces on the segments and do not account for out-of-plane motion. Calculations of moments and powers in the swing phase are not possible. In the second method for calculating joint moments, inverse dynamics based on a link-segment model approach are used.19 This method also utilizes knowledge of the magnitude and direction of the GRF and its point of application (obtained from force-plate data), as well as the location of the joint centers. This 3D method requires additional information associated with each segment (ie, the location and linear acceleration of the center of mass, the angular acceleration [obtained from the motion data], and the inertial characteristics [estimated from anthropometric relationships]),(36) The collection of valid force-plate data is not always possible, because insufficient step length may allow only one foot to have contact with a particular plate. Therefore, in small children or patients with pathology severe enough to significantly limit step length, the calculation of joint kinetics is not always possible.

To help explain the net joint moment, an example of the ankle during stance may be used. In mid-stance, the ankle dorsiflexes as the tibia rotates over the plantigrade foot. The line of action of the resultant GRF passes in front of the ankle joint center, which will tend to dorsiflex the ankle. The body's response is a net plantar-flexor moment produced by the ankle plantar flexors that counteracts the external force producing dorsiflexion. The net plantar-flexor moment increases as the center of pressure (ie, point of application of the resultant GRF on the foot) moves more distally along the plantar aspect of the foot (Fig. 3).

The net muscle moment is then used to calculate the net muscle power by multiplying the moment by the joint angular velocity.19 Power is defined as the work performed per unit of time and may be used to document the net rate of energy absorption or generation of the muscles. During normal gait, the joint power may be associated with a type of muscular contraction. For example, power generation may he associated with a concentric muscular contraction, and power absorption may be associated with an eccentric muscular contraction. Using the same example of the ankle in mid-stance, we know that a net plantar-flexor muscle moment occurs during this portion of the gait cycle. In mid-stance, the plantar flexors prevent excessive dorsiflexion (ie, the forward motion of the tibia over the plantigrade foot). The plantar flexors, therefore, would be eccentrically contracting, resulting in power absorption. In terminal stance, the plantar flexors reverse the relative motion of the ankle and plantar flexion occurs. The plantar flexors, therefore, are concentrically contracting, resulting in power generation. On a plot, power generation is typically expressed as positive and absorption is typically expressed as negative. A complete explanation of normal 3D joint kinematics and kinetics has been published previously.(21)

In order for forward progression to occur during gait, power must be generated that is consistent with the direction of forward progression. These power generators are the hip extensors during loading response, the plantar flexors in terminal stance, and the hip flexors in terminal stance/ initial swing.(37) Kinetic analysis has revealed those muscle groups that are the primary contributors to forward progression during gait. Identification of these power bursts has provided useful information that can be applied to treatment planning for patients with pathology,(9)

The clinical value of joint kinetic assessment has yet to be fully explored, because routine collection of joint kinetic data in clinical settings has only recently begun. Although adult joint kinetic data have been available(38) for some time, pediatric kinetic normative data were not available until recently. The clinical utility of kinetic information may be in the assessment of the ability of the muscles to generate power during ambulation. For example, if a particular group of muscles were found to be generating the majority of power during ambulation, any treatment that might weaken the muscle group or render it inefficient may not be beneficial for the patient. Analysis of the moments of force acting on a particular joint may also be useful in treatment planning. For example, an extended or hyperextended knee in stance can only be described in terms of degrees of motion by kinematic analysis, whereas the associated joint-moment assessment provided by kinetic analysis quantifies the net force acting about the joint (ie, an exaggerated net flexor moment may be present). The severity of the flexor moment may alert the clinician that treatment may be indicated for long-term protection of ligaments and capsular structure. In the assessment of orthoses, joint kinetics have also provided useful information for determining the effectiveness of a particular brace (see case 2 in "The Use of Gait Analysis in Clinical Decision Making" section). Further application of kinetic analysis in the evaluation of surgical intervention is needed before the utility of these measures in preoperative decision making is known.

It is possible that only through the evaluation of joint kinetics postoperatively will we better understand the inconsistencies in treatment outcome in children with CP.

Energy Expenditure

The determination of energy expenditure in gait has been undertaken since the 1950s. Ralston-39 first studied adult subjects without gait pathologies. Later, the energgy cost of gait in adult disabled populations was studied.(40-43) The necessary analyses have been used much less often in children with pathological gaits.(44-47) With the measurement systems described, we can evaluate whether a child's walking pattern is more normal following a prescribed treatment. if the energy required for walking is greater or unchanged after treatment, however, the usefulness of the intervention that generated the more normal walking pattern should be questioned.

One method of estimating energy expenditure requires the calculation of the mechanical work required for walking. Mechanical work can be calculated by using internal and external loads (kinetics) on the segments or by using kinematic data coupled with anthropometric measurements.(9,19) Using this method, the energy requirements of the individual joints can be calculated. The use of loads to judge work in pathological gait does not, however, measure the body's ability to efficiently respond to external loads. For example, the co-contraction frequently noted in patients with spasticity is unaccounted for when evaluating energy costs through the calculations of mechanical energy alone.

Another method for calculating energy expenditure is through oxygen consumption measures. This method may provide more information with respect to the child's neurologic involvement, as the total end product of the metabolic costs are estimated. Previous studies(44,47) have shown an increase in oxygen uptake in subjects with pathology as compared with subjects without pathology.

The Use of Gait Analysis in Clinical Decision Making

The following are examples of how gait analysis information can be used in clinical decision making. All of the analyses were performed in the Gait Analysis Laboratory at NCH and included the following: videotaping; clinical assessment; surface electromyography; 3D kinematic analysis; and, in case 2, 3D kinetic analysis.

Case 1: Orthopedic Surgery

Subject I was an 11-year-5-month-old girl with spastic quadriplegia secondary to CP when she was evaluated prior to orthopedic surgery. She was born prematurely at 6 1/2 months gestation, weighing 1.13 kg (2.5 lb). She began walking with a walker after 2 years of age. Previous lower-extremity surgery included bilateral subtalar arthrodeses at 5 years of age. At the time of the preoperative gait analysis, she ambulated with the use of an anterior rolling walker and bilateral hinged ankle-foot orthoses (AFOs) with free dorsiflexion and a 0-degree plantar-flexion stop. Her endurance was limited, requiring the use of a stroller for long distances. The kinematic information was collected using hand-held assistance, because her walker interfered with marker identification. This was accomplished by two clinicians on either side of the subject providing upper-extremity support in the same manner as an assistive device. Care was taken during hand-held assistance not to influence her walking velocity or facilitate changes in posture by guiding her or providing additional input through the upper extremities other than weight bearing.

Preoperative gait analysis. Pertinent clinical information is presented in Table 1. Moderate spasticity (grade 2 on a modified Ashworth scale(48)) was noted in the quadriceps femoris, adductor, hamstring, and triceps surae muscles. The subject's barefoot ambulation, as observed, consisted of a crouched pattern at the hips and knees, with apparent bilateral lower-extremity internal rotation, limited sagittal-plane motion of the hips and knees, a foot-flat pattern of initial contact on the left, and a toe-to-toe pattern on the right. When the orthoses were worn, she had increased difficulty clearing her feet in the swing phase.

Her EMGs showed continuous activity of the quadriceps femoris muscles throughout the gait cycle, prolonged activity of the hamstring muscles in stance and premature activity in mid-swing, and inappropriate activity of the gastrocnemius/soleus muscle complex in initial stance and terminal swing (Fig. 4).

Figure 5 is a graphic representation of the joint angles over one gait cycle. The right and left sides are plotted for comparison of symmetry between sides. This graph can also be compared with the normal curve shown in Figure 2. Subject l's major gait deviations were:

1. Asymmetrical pelvic obliquity. In the coronal plane, the right side of the pelvis was in exaggerated depression in terminal stance. This result may have been due to a functionally shorter right lower extremity that lacks extension at the hip and knee during this portion of the gait cycle.

2. Excessive anterior pelvic tilt. Hip flexor tightness and spasticity, along with hip extensor and abdominal weakness, may have been primary contributors to the increased anterior pelvic tilt.

3. Retracted right hemipelvis. In the transverse plane, a large asymmetrical pelvic rotation was evident, with the right pelvis retracted throughout the gait cycle. This result may have been due to the subject's decreased ability to protract the pelvis in terminal swing, or it may have been a compensation for the excessive common compensation.

4. Adducted right hip. in the coronal plane, persistent adduction of the right hip was noted and may have been due to adductor tightness or abductor weakness, or it may have been caused by the internally rotated hip in a crouched position.

Excessive hip flexion, in the sagittal plane, the thigh failed to extend fully in relation to the pelvis in terminal stance. Absence of full extension may have been associated with the excessive anterior pelvic tilt. When the pelvis is tipped anteriorly in stance, the relative motion of the hip is anatomical flexion. Full hip extension, therefore, is not required in terminal stance in order to orient the femur posterior to an absolute vertical position.

6. Right hip internal rotation. In the transverse plane, excessive right hip internal rotation was evident and was likely due to the excessive femoral anteversion on that side.

7. Decreased overall knee motion. In stance, greater-than-normal flexion was noted from initial contact to mid-stance. The greater knee flexion may have been due to hamstring muscle tightness and spasticity, gastrocnemius muscle activity pulling at the proximal end, and 10-degree knee flexion contractures. During the swing phase, the decreased knee flexion may have resulted from the abnormal continuous activity of the quadriceps femoris muscle, as is suggested by the EMG recordings (Fig. 5).

8. Limited dorsiflexion. Less-than-normal dorsiflexion was noted in stance bilaterally (greater on the right side than on the left side) and in swing on the right side. Inadequate dorsiflexion in stance or swing may have been due to gastrocnemius muscle spasticity and decreased anterior tibialis muscle control.

9. Internal left-foot progression. Because foot rotation was relative to the direction of forward progression, this rotation could have originated from any segment above the foot or from the foot itself It appeared that the subject's internal left-foot progression originated from the internal or protracted left side of the pelvis.

Recommendations/surgery. Based on clinical and gait analysis information, the following recommendations were made: right femoral derotational osteotomy to correct for excessive femoral anteversion and hip internal rotation during gait; right psoas major muscle intramuscular tenotomy to decrease anterior pelvic tilt and improve hip extension in terminal stance; a right adductor tenotomy to improve abduction in stance and swing; bilateral semimembranosus and gracilis muscle lengthenings to improve knee extension at initial contact, mid-stance, and terminal swing; bilateral distal semitendinosus muscle transfers to the distal femur to decrease flexion while trying to preserve hip extension and thus decrease the anterior pelvic tilt preliminary results, however, do not substantiate this theory); bilateral rectus femoris muscle transfers to the sartorius muscle to maintain knee flexion in swing following hamstring muscle lengthenings(7) and finally lengthening of the fascia overlying the right gastrocnemius muscle to increase its length in the stance and swing phases. In addition, at the time of surgery, the left psoas and left adductor muscles were assessed to be tight when the child was under general anesthesia. These muscles were also lengthened.

Postoperative care. Following surgery, the child was fitted for new hinged AFOs with posterior check straps to control the amount of dorsiflexion allowed in stance during postoperative gait training. A short leg cast was placed on the right leg to protect the fascial lengthening for 3 to 4 weeks postoperatively. No other lower-extremity casting was used. Physical therapy began on the second postoperative day and included gentle passive-range-of-motion exercises and instruction in a home program of range-of-motion (ROM) and positioning exercises. Precautions included no weight bearing on the right lower extremity because of the femoral osteotomy. To protect the semitendinosus muscle transfer for 3 to 4 weeks, no hip flexion beyond 45 degrees was allowed. Careful handling during assisted transfers and slow, gentle ROM exercises were used to avoid quadriceps femoris muscle reflex activity and thus protect the rectus femoris muscle transfer. Weight bearing and progressive strengthening of the lower-extremity musculature were permitted when adequate healing of the osteotomy site was noted by radiographs. Weight bearing began within the 4- to 6-week postoperative period typical for children undergoing similar surgery.

Postoperative gait analysis. One year following surgery, the subject returned to the Gait Analysis Laboratory for a routine postoperative evaluation. The preoperative and postoperative clinical measurements are presented for comparison in Table 1. Figure 6 is representative of one gait cycle during barefoot ambulation from the postoperative gait analysis. The kinematic patterns in this figure can be compared with the preoperative curves shown in Figure 5.

Figure 7 is representative of one gait cycle with the use of the orthoses and can be compared with the barefoot ambulation in Figure 6.

The subject's posture was determined by visual assessment to be more erect postoperatively, and she reported improved endurance during ambulation. The pelvic obliquity, right hip adduction, and crouched pattern at the hips and knees were eliminated. More normal transverse-plane rotation was noted, with improved overall sagittal-plane motion of the knees. For pelvic stability, she still required strengthening of the lower extremities, especially for the hip extensors and abductors, and hip flexor strength needed to be increased for clearance in swing. Improved hamstring, gastrocnemius, and quadriceps femoris muscle control may help decrease the subject's genu recurvatum. improved hip external rotation is needed for gait, and improved knee flexion is needed for mobility skills such as transfers and stair climbing. Finally, a heel shim or resetting the plantar-flexion stop of the hinged AFO to stop in 3 to 5 degrees of dorsiflexion may be helpful in controlling the residual genu recurvatum.

Summary. In this example, the gait analysis helped to identify the primary problems in all three planes of motion. Although visual observation appeared to demonstrate bilateral hip internal rotation, the subject's primary problem was excessive right femoral anteversion that led to the compensatory rotation in her pelvis and apparent left lower-extremity internal rotation. In the sagittal plane, her hamstring and quadriceps femoris muscles were limiting the sagittal-plane motion of her knees along with her plantar flexors, which were abnormally active in the swing phase and during loading response. Although the range of dorsiflexion was the same for both ankles, only the right side demonstrated shortness (ie, limited dorsiflexion) in stance and therefore warranted lengthening. in the coronal plane, the tightness of the right adductors was producing an adducted posture at the hip. When the child was under general anesthesia, however, the surgeon felt the adductors were sufficiently tight to warrant lengthening bilaterally.

Case 2:

Orthotic Decision Making

History/present Status. Subject 2 was a 17-year-old boy with the diagnosis of asymmetrical spastic diplegia secondary to CP (greater on the left side than on the right side) at the time of his gait analysis. He was born full-term without any reported complications, weighing 2.7 kg (6 lb). Developmental milestones were delayed, and he did not begin walking until 6 years of age. Approximately 1 year previous to this analysis, rear-entry hinged floor-reaction orthoses were prescribed (Fig. 8) to decrease the flexion of his knees in stance and allow plantar flexion to occur during loading response and terminal stance.

Subject 2's pertinent clinical measurements are summarized in Table 2. The pelvis in the coronal plane was within normal limits (Fig. 9). In the sagittal plane, the baseline degree of pelvic tilt was within normal limits, vet modulated in an abnormal pattern with two phases of exaggerated anterior pelvic tilt. This pattern is typical in patients with spasticity who have poor dissociation between the pelvis and femur. In the transverse plane, a mild asymmetrical pelvic rotation was noted, with the left side retracted compared with the right side. in the coronal plane, the right hip was slightly abducted in swing and early stance. In the sagittal plane, the hips showed limited overall ROM and lacked full extension in terminal stance. At the knees, the subject had difficulty with extension at initial contact, mid-stance, and terminal swing. At the ankles in the sagittal plane, he demonstrated excessive dorsiflexion in the stance and swing phases on the right side. On both sides, he lacked the normal plantar flexion that occurs during loading response as the forefoot is lowered to the floor. In the transverse plane, the left foot rotation was slightly more external than normal. When the orthoses were worn (Fig. 10), there was little change in the kinematic curves at the pelvis, hip, and knee. Both ankles were held in 0 to 10 degrees of dorsiflexion through the gait cycle, and the foot rotations were in a more neutral position.

Temporal and stride variables. Step lengths were 104% of normal (56 cm) on the right and 90% of normal on the left. Stride length was 96% of normal (1 13 cm). Cadence was 79% of normal (128 steps/min), and walking velocity was 76% of normal (120 cm/s).(18) When the orthoses were worn, these values remained relatively unchanged except for cadence and walking velocity, which were both reduced to 59% of normal.

Kinetics. During barefoot ambulation, the hip moments and powers were essentially within normal limits and therefore are not reported. At the knees, a prolonged extensor moment was evident in stance as a result of the persistent flexion of the knees (Fig. 11). With the orthoses, there were no changes in the knee kinetic patterns (Fig. 12). That is, there was a minimal reduction in the excessive extensor moment throughout stance. The minimal reduction in extensor moment may be due to the residual knee flexion contractures that interfered with the use of the GRF transmitted through the brace to the knee.

At the left ankle during barefoot ambulation, no dorsiflexor moment was present during loading response, indicating a lack of heel contact at initial contact. Even though the magnitude of the ankle power generation in terminal stance was less than normal, a distinct power generation was evident bilaterally. When the braces were worn, however, this power generation was absent. it is possible that the braces may not have allowed power to be generated at the ankle because of the position of the ankles in the braces. The subject may have had more difficulty generating power from 0 degrees of dorsiflexion into plantar flexion, as opposed to from approximately 10 to 15 degrees of dorsiflexion into plantar flexion. During gait, the ankle is normally in a position of dorsiflexion at the end of mid-stance just prior to the initiation of this power burst. Plantar-flexor activity may be triggered by this stretch. With the ankle held at 0 degrees, the plantar flexors may he at a relative disadvantage to generate power for push-off. The 0-degree dorsiflexion stop, however, is needed in order to provide a plantar-flexion/knee extension couple at the knee in stance. It is also possible that the relative decrease in walking velocity when wearing the orthoses was a result of the decreased power generation at both ankles, as power amplitude is related to walking Velocity.(38)

Summary. The orthoses did not provide adequate control of the subject's knee flexion in stance. The hinged ankle joint did not provide any additional benefits for which it was designed (ie, plantar flexion during loading response and plantar flexion in terminal stance for push-off). One might argue that a solid-ankle floor-reaction orthosis would provide the same results during gait, yet with decreased bulk at the ankle joint and reduced possibility for mechanical failure.

The evaluation of the effectiveness of this orthosis could only be accomplished through computerized gait analysis. Specifically, the kinetic analysis showed that the brace was not functioning at the ankle in the manner for which it was designed.

Criteria for Referral to a Gait Analysis Laboratory

The most common use of gait analysis is in the treatment decision-making process for children with neuromuscular disorders such as CP.(3-7) As mentioned previously, walking patterns in these children are complex and usually consist of a combination of primary problems and adaptations. Gait analysis will provide objective documentation of the patients' gait pattern and provide more insight into the causes of gait abnormalities. Usually, a referral for gait analysis is made when all methods of conservative treatment have been tried and surgical options are being considered. This typically occurs after the child has reached a plateau in terms of improvement in ambulation or when orthopedic concerns (eg, hip subluxation) necessitate treatment. Multilevel surgical procedures now allow all possible areas of dysfunction to be treated during one surgical session. Gait analysis greatly helps to identify areas of dysfunction that are not readily identified by visual and clinical assessment. With this new approach to treatment, it is possible that the "staging" of single surgeries on a yearly basis, as described by Bleck,(27) may be eliminated. With gait analysis used as a preoperative tool, the child with CP may require only one surgical treatment during the growing years. We believe, therefore, that a complete gait analysis is beneficial to any ambulatory child with a neuromuscular disorder who is being considered for orthopedic surgery.

There are, however, other factors to consider when referring a child for gait analysis. The child must be somewhat ambulatory, with or without assistive devices, for a minimum of 10 consecutive steps. A minimum height may be required (at NCH, children must be 101.6 cm [40 in] tall) in order to position sufficient markers for kinematic analysis on the child and still have a sufficient intermarker distance so that each marker can be identified. The child must be able to follow simple directions and tolerate the placement of markers and electrodes on the skin. The level of cooperation may influence the child's testing ability, given the length of a typical gait analysis. The length of a gait analysis may vary from patient to patient, depending on the number of measurements taken. if a child uses orthoses, tests may need to be completed both barefoot and with the orthoses, if the clinical question concerns brace wear. If collection of force-plate data is possible, numerous trials may be necessary to obtain sufficient data for variability comparisons, which would also increase test length. Although cooperation is typically not a major problem with the use of toys as distractions, in some cases of severe cognitive impairment, cooperation may be an issue of concern.

Gait analysis may not be appropriate as a tool for obtaining "baseline" measurements of walking ability. Although baseline measurements would provide useful information for some progressive conditions, the benefit to the patient is not clear. Gait analysis as a research tool may be more appropriate in these situations.(49) Another valuable function of gait analysis is in the assessment of surgical intervention.(7) Routine analyses of postoperative status provide the clinician with more objective information to evaluate the effect of the treatment. In general, a gait analysis may be appropriate for any patient who has a gait abnormality for which the cause is unclear or treatment decisions are complex.

Because of the limited number of clinical gait laboratories, most facilities allow referral of patients. At NCH, where gait analysis is a part of the surgical decision-making process, referral is made typically by an orthopedist. Referrals from other physicians are also accepted. A therapist may initiate a referral process, but the final referral must be made by the patient's physician. Patients who do not have an orthopedist may be initially referred to the appropriate outpatient clinic based on diagnosis, after which a referral to the gait laboratory would be made, if appropriate. Once a patient is scheduled for a gait analysis, input from the referring physician and physical therapist regarding the patient's present status and any other information that may influence the surgical decision-making process is requested.

Summary

Computerized gait analysis provides the tools necessary to evaluate both normal and pathological gait. Through the study of normal gait, we are better able to understand pathological gait and to prescribe more effective treatments. Recent advances in technology have allowed us to measure gait variables more precisely and in a timely manner. There is no single method of measurement, however, that provides a complete analysis of gait. We believe a complete analysis should include videotaping, clinical evaluation, EMG, 3D joint kinematics, 3D joint kinetics, and the evaluation of energy expenditure. Sophisticated clinical gait analysis requires the integration of many methods of analysis to arrive at a more complete assessment of a child's gait pattern. Proper interpretation and integration of the information obtained from sophisticated gait analysis measures is possible with a basic knowledge of how the information is collected, processed, and calculated. The purpose of this article was to review these various components of modern gait analysis and demonstrate the applications of this information in clinical decision making for the ambulatory child with neurological involvement. The objective documentation of gait also allows for the evaluation of treatment outcomes and subsequently aids in the development of new ideas.

References

1 Simon SR, Deutsh SD, Nuzzo RM, et al. Genu recurvatum in spastic cerebral palsy. j Bone Joint Surg [Am]. 1978;60: 2-894.

2 Sutherland DH, Cooper L. The pathomechanics of progressive crouch gait in spastic diplegia. Orthop Clin North Am. 1978;9: 143-154.

3 Perry J, Hoffer MM, Goeovan P, et al. Gait analysis of the triceps surae in cerebral palsy J Bone Join Surg[AM]. 1974;56:511-520.

4 Perry J, Hoffer MM. Preoperative and postoperative dynamic electromyography as an aid in planning tendon transfers in children with cerebral palsy. Bone joint Kurg [Am]. 1977;56:531-537

5 Gage JR, Fabian D, Hicks R, Tashman S. Preand postoperative gait analysis in patients with spastic diplegia: a preliminary report. J Pediatr Orthop. 1984;4:715-725

6 DeLuca PA Gait analysis in the treatment of the ambulatory child with cerebral palsy. Clin Orthop. 1991;264:65-75.

7 Gage Jr. Surgical treatment of knee dysfunction in cerebral palsy. Clin Orthop. 1990;253: 45-54.

8 Hicks R, Durinick N, Gage JR. Differentiation of idiopathic toe--walking and cerebral palsy. J Pediatric Orthop. 1988;8:160-163.

9 Olney SJ, MacPhail HEA, Hedden DM, Boyce WF. Work and power in hemiplegic cerebral palsy gait. Phys Ther. 1990;70:431-438.

10 Sutherland DH, Wyatt MP, Biden EN. Use of the motion analysis laboratory. in: Goldsmith's Practice of Surgen,. Philadelphia, Pa: Harper & Row, Publishers Inc; 1985: chap 8.

11 Gage JR, Outipuu S. Surgical intertervention in the correction of primary and secondary gait abnormalities. In: Adaptability of Humans Gait: Implications for the Control of Locomotion. Amsterdam, the Netherlands: Elsevier Science Publishers BV; 1991:359-385.

12 Krebs DE, Edelstein JE, Fishman S. Reliability of observational kinematic gait analysis. Phys Ther. 1985;65:1027-1033.

13 Pathokinesiologv Department and Physical Therapy Department. Observational Gait Analysis Handbook. Downey, Calif. The Professional Staff Association, Rancho Los Amigos Medical Center; 1989.

14 Winter DA. Concerning the scientific basis for the diagnosis of pathological gait and for rehabilitation protocols. Physiotherapy Canada. 1985;37:245-252.

15 Young CC, Rose SE, Biden EN, et al. The effect of surface and internal electrodes on the gait of children with cerebral palsy spastic diplegic type. J Orthop Res. 1989;7:732-737

16 Stuberg WA, Colerick VI, Blanke Dj, Bruce W. Comparison of clinical gait analysis method using videography and temporal-distance measures with 16-mm cinematography. Phys Ther. 1988;68:1211-1225,

17 Davis RB, Ounpuu S, Tyburski Dj, DeLuca PA. A comparison of 2D and 3D techniques for the determination of joint rotation angles. Proceedings of the Internal Symposium on 3-D Analysis of Human Movement. Montreal, Quebec, Canada, 1991.

18 Sutherland DH, Olshen RA, Biden EN, Wyatt MP. The Development of Mature Walking. don, England: MacKeith Press; 1988.

19 Winter DA. Biomechanics and Motor Control of Normal Human Movement. 2nd ed. New York, NY: John Wilevy & Sons inc; 1990.

20 Lippold OCJ. The relationship between integrated action potentials in a human muscle and its isometric tension. J Physiol (Lond). 1952;177:492-499.

21 Vredenbregt J, Rau G. Surface electromyography in relation to force, muscle length and endurance. In: Desmedt JE. Delelopments in Electronivography and Clinical Neurophysiology. Basel, Switzerland: S Karger AG, Medical and Scientific Publishers; 1973;1:607-622.

22 Basmajian JV, DeLuca CJ. Muscles Alive: Their Functions Revealed by, Electronlyography. Baltimore, Md: Williams & Wilkins; 1985.

23 Yang JF, Winter DA. Electromyographic normalization methods: improving their sensitivity as a diagnostic tool. Arch Phys Med Rehabil. 1984;65:517-521.

24 Ounpuu S, Winter DA. Bilateral electromyographical analysis of the lower limbs during walking in normal adults. Electroencepbalogr Clin Neurophysiol. 1989;72:429-438.

25 Yang JF, Winter DA. Electromyographic reliability in maximal and submaximal isometric contractions. Arch Phys Med Rebabil. 1983;64:417-420.

26 Winter DA, Yack HJ. EMG profiles during normal human walking: stride-to-stride and intersubject variability. Electroencephalogr Clin Neurophysiol. 1987;67:402-411.

27 Bleck EE. Orthopaedic Management in Cerebral Palsy. Philadelphia, Pa: MacKeith Press; 1987.

28 Ounpuu S, Gagejr, Davis RB. Three-dimensional lower extremity joint kinetics in normal pediatric gait. J Pediatr Orthop. 1991;11:341-349.

29 Tata JA, Peat M. Electromyographic characteristics of locomotion in normal children. Physiotherapy Canada. 1987;39:167-175.

30 Kadaba MP, Wooten ME, Gainey J, et al. Repeatability of phasic muscle activity: performance of surface and intramuscular wire electrodes in gait analysis. J Orthop Res. 1985;3: 350-359.

31 Komi PV, Buskirk ER. Reproducibility of electromyographic measures with inserted wire electrodes and surface electrodes. Electromyography. 1970;10:357-367.

32 Perry J, Easterday CS, Antonelli Dj. Surface versus intramuscular electrodes for electromyography of superficial and deep muscles. Phys Ther. 1981;61:7-15 33 Whittle M. Gait Analysis: An Introduction. Oxford, England: Butterworth & Co (Publishers) Ltd; 1991.

34 Winter DA. Kinematic and kinetic patterns in human gait: variability and compensating factors. Human Movement Science. 1984; 3:51-76.

35 Davis RB. Clinical gait analysis. Eng Med Biol. 1988;17:35-40.

36 Dempster WT, Grabel WC, Felts WJL. The anthropometry of manual work space for the seated subject. Am J Phys Anthropol. 1959; 17:289-317.

37 Winter DA. The Biomechanics and Motor Control of Human Gait: Waterloo, Ontario, Canada: University of Waterloo Press; 1987

38 Winter DA. Energy generation and absorption at the ankle and knee during fast, natural and slow cadences. Clin Orthop. 1983;174: 147-154.

39 Ralston HJ. Energy-speed relation and optimal speed during level walking. Int Z Agnew Physiol. 1958;17:277-283.

40 Bard G. Energy expenditure of hemiplegic subjects during walking. Arch Phys Med Rehabil. 1963;44:368-370.

41 Bard G, Ralston HJ. measurement of energy expenditure during ambulation with special reference to evaluation of assistive devices. Arch Phys Med Rehabil. 1959;40:415-420.

42 Corcoran PJ, Brengelmann GL. Oxygen uptake in normal and handicapped subjects in relation to speed of walking beside velocity-controlled cart. Arch Phys Med Rehabil. 1970;51:78-87.

43 Gordon EE, Vanderwalde H. Energy requirements in paraplegic ambulation. Arch Phys Med Rehabil. 1956;37:276-285.

44 Campbell J, Ball J. Energetics of walking in cerebral palsy. Orthop Clin North Am. 1978; 9:358-360.

45 Dahlback GO, Norlin R. The effect of corrective surgery on energy expenditure during ambulation in children with cerebral palsy. Eur J Appl physiol. 1985;54:67-70,

46 Lough LK, Nielsen DH. Ambulation of children with myelomeningocele: parapodium versus parapodium with ORLAU swivel modification. Dei, Med Child Neurol. 1986;28: 489-497.

47 Rose JR, Gamble JG, Medeiros J, et al. Energy cost of walking in normal children and those with cerebral palsy: comparison of heart rate and oxygen uptake. J Pediatr Orthop. 1989;9:276-279.

48 Bohannon RW, Smith MB. Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther. 1987;67:206-207.

49 Winters TF, Hicks R, Gage JR Gait patterns in spastic hemiplegia in children and young adults. J Bone Joint Surg [Am]. 1987;69:4337-4411.