People with spinal cord injury at the C5/C6 motor complete level typically have relatively well-preserved biceps function but minimal or no voluntary control of triceps. The authors previously have shown the lack of voluntary control of triceps results in deficiencies in speed and accuracy of elbow movements as well as reduction in reachable workspace. The authors also have shown these deficiencies can be corrected by the addition of constant extensor torque at the elbow. The purpose of this article is to describe a prototype constant-torque elbow orthosis and illustrate the improvements in function it affords people with C5/C6 tetraplegia.
Spinal cord injury causes the loss of voluntary control of muscles and varying levels of decreased mobility and dexterity depending on the level and completeness of injury. People with injuries at the C5 or C6 motor complete level whose biceps function is relatively well-preserved often have little or no voluntary control of their triceps. A flexorextensor muscle imbalance such as this, combined with even a small amount of biceps spasticity, can lead to the development of a flexed and supinated forearm.
A number of orthoses have been designed to prevent and help correct elbow-flexion postures or contractures. For example, static orthoses use a turnbuckle or other means to permit the elbow to be locked at a preset angle of extension that gradually is increased as contractures are reversed (1,2). Dynamic plastic (3) and spring-loaded orthoses such as Wire-Foam a (4) and Dynasplint(r) (5) provide a continuous, low-intensity extension force to reverse flexion contractures. The Wire-Foam orthosis obtains force from a pair of coil springs located on the lateral and medial sides of the axis of elbow rotation. The Dynasplint consists of two stainless steel struts placed laterally and medially on the upper extremity, with adjustable compression-coil springs incorporated within the distal forearm struts.
Abrahams et al. (6) have combined a pair of medial and lateral coil springs with geared hinge joints to produce extension force and spring wire coupled to the distal forearm cuff to provide pronation force. Itzkovich et al. (7) have described a dynamic orthosis comprised of arm and forearm braces made of polyethylene and joined at the elbow using flail arm hinges with one side locking. Extension force is provided by a rubber ring on the posterior of each brace. Static activities can be accommodated by locking the lateral hinge.
Another problem related to weak or absent triceps in C5/C6 subjects is the deterioration of control of upper-limb motion even when contractures at the elbow are absent. Controlling the arm without the triceps muscle is like driving a car without brakes: It is difficult to stop moving. For example, one patient complained about hitting himself with his fork while attempting to feed himself. In general, without use of the triceps (which is the primary elbow extensor), tetraplegic subjects face limits not only to their range of accessible space but also to their ability to perform tasks in which the biceps muscle does the work.
The authors recently showed tetraplegic subjects with weak or absent triceps performed elbow-flexion movements substantially slower than did control subjects and made larger overshoot errors when making flexion movements to a target (8). When a torque motor was used to aid the weak triceps by providing a constant extensor torque of 2.5 Nm, which is approximately 6 percent of maximum "normal" triceps strength of 39 Nm on the authors' apparatus (8), patients performed elbow flexions more quickly and with accuracy indistinguishable from that of the control subjects.
Improving upper-limb function of a tetraplegic person is important because it allows an injured individual to lead a more independent life. However, to date, the application of elbow orthoses to tetraplegic subjects largely has been for the prevention or correction of elbow contractures. The broader problem of improving tetraplegics' control of arm movements has not been addressed. All commercially available spring-driven dynamic elbow orthoses use a linear spring, allowing the extension torque to change with elbow angle, meaning extension torque is maximal when the elbow is most flexed and minimal when most extended.
The authors have developed an orthosis that provides constant extensor torque at the elbow. The design is based on the authors' studies showing an improvement of arm function in C5/C6 tetraplegia with the use of a torque motor (8,9). The new orthosis differs from existing orthoses in two significant ways.
First, it allows constant extension torque during activities performed close to the mouth (when the arm is flexed) and when reaching overhead (when the arm is fully extended). Since torque does not increase with flexion, fatigue of the biceps when working with flexed postures is limited; yet sufficient torque is available to fully extend the arm in any position. In contrast, a linear spring does not provide enough force at "full extension" to allow such extension overhead.
Second, subjects are more comfortable and more adept when controlling arm movements against a constant torque than against the progressive torque from a linear spring or rubber band (9).
Two alternative methods of restoring triceps function that do not involve orthoses are posterior-deltoid-to-triceps tendon transfer surgery (10) and functional electrical stimulation (FES) (11) Tendon transfers are an attractive option because they are completely internal and sometimes can provide sufficient strength for lifts for skin pressure relief. However, tendon transfer is no feasible if the posterior deltoid has in sufficient strength, and the procedure does require surgery plus a substantial period of postsurgical immobilization and rehabilitation.
FES is effective only in subjects with out extensive triceps denervation. Because FES of a muscle relies on electrical activation of the peripheral motor nerve, the muscle cannot be electrically activated if death of its spinal motor neurons (denervation) has led to loss of peripheral motor axons. The FF5 approach to restoration of triceps function is cosmetically advantageous as it avoids the use of a bulky orthosis; however, this approach requires the application of skin electrodes, tolerance of sensations associated with stimulation and conditioning of the triceps muscle to reduce the problem of muscle fatigue. An orthosis remains an option for patients who want improved function but for whom the alternatives are unfeasible or undesirable.
The overall objective of this study is to design and test a purely mechanical orthosis that improves arm motion in tetraplegic subjects who have injuries at the C5/C6 level. The purposes of this article are to: 1) present a prototype of the constant-torque elbow orthosis, and 2) demonstrate improvement of arm function in C5/C6 subjects using the orthosis.
The prototype one-hinge orthosis shown in Figure 1 consists of four cuffs (see Figure 1, c1-c4 ) and one hinge incorporating a spring (see Figure 1,s ) on the lateral side of the elbow. The cuffs are fabricated from two layers of carbon-fiber lamination braid C designed for light weight and high stiffness and lined with 3/8-inch blue T-Foam for customized fitness and comfort. Having a pair of cuffs both proximal and distal to the elbow allows for uniform distribution of torque generated at the elbow joint and helps keep the orthosis in place. The hinge arms (see Figure 1, h ) are made of stainless steel tubes 7 mm in diameter and 0.11 mm thick.
The hinge on the lateral side of the elbow incorporates a spring mechanism (see Figure 2 ). A power spring C (see Figure 2, a ) is used to provide torque in the direction of elbow extension. The inner end of the spring is attached to a ball-bearing mechanism to allow for low-friction rotation. A lightweight spring cage (see Figure 2, b ) is provided to hold the spring in place, and a mechanical stop (see Figure 2, c ) prevents extension of the elbow beyond the anatomic limit. The hinge and tubes are welded together to avoid unnecessary connecting parts that would add weight, the result of which can be fatiguing to subjects with proximal weakness in the upper limb. The weight of the authors' prototype is about 600 g.
Torque can be changed in this prototype by winding or unwinding the power spring using a ratchet, much as one would wind a clock. A removable handle (see Figure 2, d ) is provided for convenience. When the orthosis is fitted to the arm, the spring is prewound to an appropriate torque, usually to 2 Nm (four full turns). At this level, users can extend their upper limbs overhead. If the spring is fully wound, it will produce constant torque (± 10 percent) over the full range of elbow flexion, 0-150 degrees. Torque can be decreased (see Figure 2, e) when donning or doffing the orthosis, making those actions much easier, then restored once the orthosis is in place. A second advantage of this ratchet is it allows the user to change the spring torque at any time, increasing it to provide greater function or decreasing it to provide greater comfort.
A second spring, used to enhance pronation, also can be seen in Figure 1 . Individuals who have injuries at the C5 level retain their biceps muscle, which can act as a supinator of the wrist, but lose the ability to pronate the wrist. A few patients with injuries at the C6 level may retain some ability to pronate. The prototype orthosis assists pronation by means of a second spring (see Figure 1, w ) attached at one end to the lower hinge arm (h) and at the other to the distal forearm cuff (c4). Pronation torque is enabled by winding up the distal cuff (rotating it clockwise in the case of the right arm) around the hinge arm several times prior to locating the cuff on the wrist. As the cuff unwinds (counterclockwise) around the hinge arm, it assists pronation of the hand. The pronation torque is weak because of the small diameter of the hinge arm (7 mm) and some unavoidable misalignment of the spring axis and the pronation/supination axis of the forearm. The forearm spring is set to be approximately one-tenth as strong as the elbow spring. Although this spring did appear to facilitate pronation, this function was not evaluated quantitatively and will not be discussed further in this article.
An earlier prototype, a two-hinge orthosis (illustrated in Reference 12), has springs incorporated into hinges on both sides of the elbow. The hinge on the medial side makes the orthosis stiffer and more stable as all torques are in the plane of elbow flexion. In that model, cuffs retain rotation along the pitch axis (see Figure 1, p ), ensuring automatic positioning of the cuffs to distribute pressure evenly onto the skin. However, two users noted the medial-hinge spring interfered with manual propulsion of the wheelchair.
Modifying the prototype by removing the medial hinge, which is absent in the one-hinge prototype, makes the orthosis lighter and less intrusive. However, the modified orthosis becomes subject to twisting torques and requires more attention to orthosis rigidity. The two-hinge orthosis might be preferred by patients who use a powered wheelchair and thus do not find the medial hinge intrusive.
Part of the challenge of designing a one-hinge orthosis is ensuring proper balance of the various forces acting at the elbow so the orthosis will remain in place on the arm when the user flexes or extends his/her arm. The center of rotation of the orthotic hinge should remain aligned with the anatomic hinge of the elbow. However, an orthosis designed to provide torque to extend the arm at the elbow using only one hinge will tend to twist around the arm to straighten itself without straightening the arm. Increasing the rigidity of the orthosis minimizes the tendency to twist; however, this increased rigidity also decreases comfort.
The authors' solution to the problem of discomfort was to provide sufficient degrees of freedom in the attachment of cuffs to the hinge so the cuffs could be adjusted to provide for appropriate, equal distribution of forces as well as comfort. The first one-hinge design allowed for two rotations: one around the axis of the hinge arm, a "roll" rotation (see Figure 1, r ), and another perpendicular to it, the "pitch" rotation (see Figure 1, p).
In addition, translation (sliding) of the cuff along the hinge arm was possible. Cuff pitch was permitted using only one screw to mount the cuff to the hinge arm. Cuff roll and translation could be limited after proper adjustment by tightening set screws (see Figure 2, )(f, g) on the hinge arm to provide the best fit and performance. Although these set screws are adequate for a prototype, more robust connectors may be required for continuous clinical use.
Disallowing rotation of the cuffs along the pitch axis was necessary since this degree of freedom contributed strongly to the twisting of the orthosis. Cuff pitch was eliminated by attaching the cuffs to the hinge arms with two screws, above and below the hinge arm (see Figure 1 ). In the absence of cuff pitch, the cuffs could still be translated more proximally and distally along the hinge arm and rotated around the hinge arm. The latter adjustment was especially useful for donning or doffing the orthosis: A cuff could be rotated out of the way and then rotated back into place. After tightening the set screws to make the orthosis rigid, the balance of forces acting on the arm was surprisingly good but still not as good as in the two-hinge orthosis (12), which stayed on the arm even if no Velcro straps (see Figure 1, (v) were used.
Selection of Subjects Subjects were recruited from the outpatient population of the Spinal Cord Injury Service at the Brockton/West Roxbury VA Medical Center. Three tetraplegic men with motor-complete spinal cord injury at the C5/C6 level whose biceps were relatively functional [4/5 to 5/5 by manual muscle testing (MMT)] and whose triceps were weak (215 or less) participated in the study. Subjects' ages were 46,32 and 19 years, and times since injury were 2,9 and 0.3 years, respectively.
Individuals were excluded from the study if they had severe spasticity, very limited range of motion, unhealed bone fractures or other significant medical problems in addition to SCI. The number of subjects participating in each test is given in the description of that test.
The evaluation tests were designed to investigate improvements in: 1) controlling arm motion (tests 1 and 2), 2) expanding the reachable workspace (tests 3 and 4), and 3) functional ability to perform daily tasks (test 5).
All tests were performed in a quiet setting in the Motor Control Laboratory of the Spinal Cord Injury Service. Tests were scheduled at the convenience of the subjects and always were conducted by the same research team. Each subject was seated in his own wheelchair and performed the required task without outside help. Because the one-hinge and two-hinge orthoses discussed above produced similar functional results when adjusted to provide similar torque, the results of testing have been pooled.
Test 1: Single-Joint Speed and Accuracy This test was designed to isolate the function of the elbow flexors and extensors. Subjects performed fast elbow-flexion movements in which the biceps muscle accelerated the limb and the triceps acted as a brake to decelerate the limb and stop the arm at the intended place (13,14). To quantify improvement in controlling arm motion with use of the orthosis, rapid movements aimed at a visual target were compared with and without the orthosis on the arm (10 trials under each condition).
Subject I (age 46; two years since injury) sat in his own wheelchair with his elbow supported on a table at shoulder level and strapped to a manipulandum, allowing horizontal rotation of the forearm at the elbow but limiting any contribution of shoulder muscles to the movement. An oscilloscope in front of the subject displayed two horizontal lines: one representing the initial and then final target, and the other indicating the subject's actual arm position. The subject's task was to align his arm position with the initial target line at the beginning of each trial and then move his arm "as quickly and accurately as possible'' to the final target position when the target line shifted. Movements of 10 and 30 degrees were performed.
The angular position of the arm was recorded with a potentiometer attached to the axis of the manipulandum. The position data were sampled at 1 kHz for one second after a target presentation. Velocity and acceleration were calculated by numerical differentiation of the position data. Parameters including peak displacement, peak velocity, movement time, and duration of the accelerative and decelerative phases were evaluated off-line according to predefined criteria detailed in Reference 8. Errors in performance were measured as the difference between actual peak displacement and the intended target and were expressed as a percentage of the target amplitude. Constant error (mean percent overshoot) and variable error (standard deviation of within-subject constant errors) were evaluated.
The results of this test indicate the orthosis facilitated faster movements; average (± SD) movement times decreased (p<0.0) from 312 (± 34) ms to 24 (± 31) ms in 10-degree movements of one subject and 468 (± 49) ms to 294 (± 35) ms in 30-degree movements (see Figure 3a ). Acceleration and deceleration times also were improved (see Figure 3b ). The variability of movements (standard deviation of errors) was statistically similar with and without the orthosis (see Figure 3c ) although average overshoot (constant error) was de creased during 30-degree movements with the orthosis. However, the subject moved faster while wearing the orthosis. These faster movements, if made without the orthosis, would have resulted in considerably larger errors in overshooting the target. The improvements in motor performance with use of the orthosis are similar to those reported in previous studies using a torque motor rather than an orthosis (8,9).
Test 2: Multijoint Coordination This test examined improvement in interjoint coordination enabled by the orthosis, reflecting the improved ability of the user to compensate for the force of gravity acting on the arm. The subject was asked to draw horizontal and vertical ellipses in the air in front of him (frontal plane) with the palm of his hand. A drawing of an ellipse, with long axis 16 cm and short axis 7 cm, was shown to the subject to indicate the approximate size and shape of the ellipses he was to draw. The trajectory of the arm movement was recorded by means of two miniature accelerometers (each weighing only a few grams) mounted perpendicularly to each other and attached to the subject's hand.
Data from these accelerometers were transferred to a computer program that calculated the displacements of the movement trajectories in the frontal plane. Only subject 1 participated in this study. As can be seen in Figure 4 , use of the orthosis allowed the subject to draw less distorted vertical (see Figure 4b ) and horizontal (see Figure 4d ) ellipses in comparison to the drawings he made without the orthosis (see Figure 4 a,c), when his hand could not maintain its intended path through space because of his lack of triceps strength.
Test 3: Front Reaching The purpose of this test was to show whether wearing an orthosis would allow for an increase in the subjects' range of arm movements. Three subjects performed reaches, with and without the orthosis, to a 60-cm high, 120-cm wide "target board" (see Figure 5 ) placed on a table 45 cm in front of their chests. Targets were 4-cm diameter holes arranged in four rows of seven columns, spaced 18 cm apart. A lightweight pen was strapped to each subject's index finger.
Each subject was instructed to bring his hand from the initial resting position on the table to each target. Success was based on ability of the subject to place the pen into each hole and maintain that position. Requiring the subject to hold the hand at the intended target for a couple of seconds allowed the authors to differentiate targets that could truly be reached from targets that could only be touched by "flinging" the arm momentarily to the target.
After the drill described above, reach speed was tested. Each subject inserted the pen in each target hole, one after the other, returning the hand to the resting position on the table after each insertion. The entire series was timed, and only those holes the subjects could reach both with and without the orthosis were included in the speed test.
Results of the front-reaching test varied among subjects because a test of three-dimensional reaching requires coordinated movement and strength of several muscle groups, including shoulder muscles. For example, in lifting the arm overhead, both triceps and shoulder muscle strength are important. Since the pattern of weakness following spinal cord injury varies from person to person, subjects' performances on three-dimensional tasks are expected to reflect not only the effect of the orthosis but also differences in strength of other muscles. In contrast, during single-joint tasks (Test 1) in which motion is restricted to elbow rotation, subjects' performances more closely reflect their triceps weakness (8).
Subject 1, who had no triceps strength (0/5 by MMT) and fairly weak deltoid muscles, showed marked improvement in his ability to reach the higher and contralateral targets while wearing the orthosis (see Figure 6b) ; the targets were not accessible without it (see Figure 6a ). For example, when attempting to reach the higher targets without the aid of the orthosis, the subject's hand often collapsed onto his head. These targets became easily accessible with the aid of the orthosis. In the timed task, the subject's arm moved quickly to 15 targets (the lower three rows and inner five columns that also could be reached without orthosis) when wearing the orthosis. This task required 37 seconds without the orthosis and 30 seconds with the orthosis.
Subject 2 (age 32; nine years since injury) had no useful triceps strength (0/5 by MMT) and was limited by a 20degree flexor contracture in the tested (left) arm; nevertheless, he showed the ability to reach two additional targets (one in the highest row directly in front of him and another slightly contralateral to him in the second row from the top) while wearing the orthosis. However, contralateral reaching to the leftmost column was not restored by the orthosis because of the contracture. The timing test, which in this case involved 16 target locations, indicated the subject could move faster with the orthosis (32 seconds) than without it (42 seconds).
Subject 3 (age 19; four months since injury) had a weak triceps muscle (2+/5 by MMT) but strong deltoids and could reach all targets even without wearing the orthosis. This subject also had some control of elevation of his arm above his head but could not fully extend his arm overhead. Subject 3 might not be a suitable candidate for wearing the orthosis because he would show little functional improvement from its use.
In summary, this test showed the elbow orthosis extended subjects' reaching space, allowed faster movements and abolished arm collapse associated with placement of the hand above shoulder level.
Test 4: Side Reaching In this test, subjects were asked to extend the upper limb to the side and abduct it as high as possible both with and without the orthosis while seated in a wheelchair. The distance from the Floor to the subject's hand was measured at full abduction or when the elbow collapsed into flexion. Subject 1 increased his vertical side reach from 129 to 142 cm, subject 2 from 112 to 127 cm, and subject 3 showed no improvement.
Test 5: Propelling the Wheelchair This test, simulating wheelchair propulsion movements, was performed only by subject 1 (subject 2's contracture precluded a meaningful test, and subject 3 had full range of backward reaching). The task was to sit in the wheelchair and reach back as far as possible to contact the push rim. Without the orthosis, subject 1 could extend his arm only to a gravity-assisted 48 degrees of flexion before touching the rim. With the orthosis, he could further straighten his arm to 23 degrees of flexion, allowing a longer power stroke.
In earlier studies of the role of flexor and extensor muscles in controlling fast single-joint movements to a target (13), the authors found the extensor muscle, in this case the triceps, plays an important role in facilitating faster movements. Such basic research in neuromuscular motor control recently has led to recognition of deficits in control of arm movements of C5/C6 tetraplegic subjects who retain little or no voluntary triceps activity even in the absence of biceps spasticity or flexion contractures (8).
These findings suggest the applicability of elbow extension orthoses to the C5/C6 tetraplegic population may be far wider than anticipated. In general, orthoses designed to treat contractures at the elbow are designed to increase the elbow range of motion after therapy but are not designed to improve active upper-limb motion.
Even spring-loaded designs are not suitable for this purpose since the force of their linear springs declines in proportion to elbow extension. In overhead reaching, the extension torque provided by a linear spring would decline with arm extension to the point of inadequacy. A second problem with linear springs is they contribute to control instability (tremors) that subjects themselves can recognize (9). For these reasons, the key innovation of the present design is the use of a spring that provides constant (or nearly constant) torque over the full range of elbow movement.
The authors' goal is to develop a lightweight orthosis composed of passive mechanical elements (primarily a constant torque spring) to improve control of the upper limb. The orthosis is not intended to provide the substantially larger forces required for transfers (e.g., from bed to wheelchair) because the added bulk and weight this would require would be unacceptable for ordinary use. The orthosis also is not intended for the reversal of flexion contractures; but used regularly, it may assist in the prevention of contractures by promoting elbow movement through the full range of motion. The functional tests indicate the prototype orthosis allows for faster, better coordinated upper-limb movements and provides increased reaching capacity for people with C5/C6 tetraplegia. Although the current prototype meets the authors' goals for improved function, the design could be improved by reducing its bulk, increasing stability and cosmesis, and, ideally, eliminating dependence on outside help for donning/doffing and adjusting spring-force levels.
The authors have chosen not to seek a patent on the concept of constant extension torque so other practitioners can creatively use it in their own orthotic designs. As opposed to splints developed for temporary clinical purposes, an orthosis intended for continuous use at home must excel in terms of convenience, long-term comfort and attractiveness.
This work was supported by grants from the Paralyzed Veterans of America Spinal Cord Research Foundation. The authors are grateful to Professor Tomasz Wierzbicka for his valuable help in mechanical design of the orthosis. The authors also appreciate the ideas and manufacturing skill of machinist Steven Rudolph and Dennis Amtower, CPO.
M. MARGARET WIERZBICKA, PhD, is assistant professor of neurology at Harvard Medical School in Boston.
ALLEN W WIEGNER, PhD, is assistant professor of neurology (biomedical engineering) at Harvard Medical School and a biomedical engineer in the Spinal Cord Injury Service, West Roxbury Department of Veterans Affairs Center, Boston, MA 02132; (617) 323-7700, ext. 5387.