Motion restriction of eight knee orthoses were compared for the pathological conditions of valgus, varus and axial rotation. The eight knee orthoses included the Polyaction, Lerman, Lenox Hill, ECKO, DonJoy Analog, Pro Am and CTi. Valgus, varus and axial rotation torques were applied to the test apparatus and corresponding angular deformities were measured.
Analog and CTi provided the most valgus control, Analog the varus control and Lerman and CTi the most axial rotational control. The clinical efficacy of the knee orthoses evaluated depended upon the magnitude of the imposed load and the quality of fit.
The knee is a modified hinge joint capable of the fundamental motions of flexion and extension in addition to some rotation and gliding motion. Control of knee motion is shared by the capsular structures, intra- and extra-articular ligaments, the joint contours and the muscles with their tendons which cross the joint anteriorly and posteriorly.1,2 While sagittal plane motion (flexion/extension) consists of a large, free arc of approximately 150 degrees, motion in the coronal and transverse planes are very small and limited by ligamentous and capsular structures.1,2 In a normal and fully extended knee, axial rotation is less than 20 degrees and combined varus/valgus is less the 5 degrees. 1,3,4 As the knee is flexed towards 90 degrees, increased axial rotation to 90 degrees may be attained,1,5 while combined varus/valgus is less than 15 degrees.6 While the knee is most stable when fully extended, it is also in this position that it is most vulnerable to injury, especially from lateral/medial forces.
Injury to the knee may exist as fracture of bone or tearing or rupturing of ligaments and menisci. Understanding the role of the bony and soft tissue structure of the knee is mandatory for proper management of the posttraumatized knee.
Specific function of the knee ligaments and the consequence of their injury have been described 7,8,9as well as tests to determine the extent of each.10,11 Knee joint instability, secondary to tearing or rupturing of the soft tissue structures, is clinically described by the resulting excessive motion. Varus is an unstable knee in which the tibial plateau leans laterally ("bow-legs") and is the result of disruption of the lateral structures (mid-third capsular ligament, fibulo-collateral ligament, biceps tendon, and ilio-tibial tract, with or without abnormality of the medial meniscus).8,9 In valgus, the tibial plateau leans medially ("knock-knees") and results from disruption of the medial structures (medial capsular ligament, posterior/anterior cruciates).8,9 Axial rotation refers to the tibial plateau turning externally or internally with respect to the femur. Rotation injuries can result from almost any combination of structural damage but primarily result from anterior cruciate, medial capsular and tibial collateral ligament damage.7,9 Drawer sign refers to the tibial plateau sliding anteriorly or posteriorly with respect to the femur while the patient's knee is flexed. The drawer sign is most commonly linked to tear of either the anterior and/or posterior cruciate ligaments. However, some feel that to get a significant anterior drawer sign the medial capsular ligament must also be damaged.9,11 Lastly, recurvatum is the increased posterior angle of the leg while the patient's knee is hyper-extended. Posterior support of the knee has been found to be offered primarily by the posterior cruciate ligament, posterior capsule and tibial collateral ligament.7,9
The excessive motions allowed by an injury to the knee need to be controlled via conservative management to enable healing, or to protect the knee following surgical repair. Knee orthoses have been used as an adjunct to the management of the post-traumatized knee and designed to immobilize after surgery or injury, and/or to protect joints that were painful due to trauma or disease.12 Lately, knee orthoses have been designed to stabilize antero-lateral and antero-medial rotatory motion ,13 medio-lateral motion,14,15 and excessive flexion/hyper-extension motions.16 There also have been efforts toward preventing pain by providing positioning support of the patella via an orthosis. 17,18
Designs have been as simple as those reported by Palumbo17 and Levine18 which consist of straps which surround or support the patella to those which are customized with specific biomechanical objectives for individual patients.13'14'15
Although the literature is abundant with information on the anatomy and pathology of the knee, few studies are reported that objectively compare treatment using an orthosis versus no orthosis and even fewer compare knee orthoses. A survey was conducted to determine the outcome of collegiate football players wearing versus not wearing prophylactic knee orthoses.19 Surprisingly this survey showed a greater incidence of injury in those who wore an orthosis than those who did not, regardless of the type of knee orthosis worn. Butler, et al.20 described 13 knee orthoses commonly prescribed for patients with ligamentous laxity secondary to rheumatoid arthritis or injury. Butler commented on the "potential function" of these knee orthoses and was critical of the short moment arms, the application of biomechanical forces on compliant soft tissues and suspension. Butler created a prescription guide by subjectively evaluating warmth, comfort, protection against blows and limitation of various deforming motions. Motion restriction was graded as slight, fair or good. A prescriber with specific and subjective treatment goals could use Butler's guide to zero-in on one of the 13 knee orthoses described. Knutzen, et al. ,21 compared an Ace elastic support (Becton Dickinson Co., Rochelle Park, NJ) to the Lenox Hill derotation knee orthosis (Lenox Hill Brace Shop, 2245 B. 84th St., New York, NY) on runners for flexion, axial rotation and varus/ valgus using electrogoniometric evaluation of knee joint kinematics. The Lenox Hill derotation knee orthosis restrained axial rotation of surgically repaired runners' knees approximately 2.3 degrees and the Ace support had no effect. Basset evaluated the effectiveness of the Lenox Hill orthosis in controlling anterolateral rotatory instability and combined anteromedial-anterolateral rotatory instability in 36 patients using the standard clinical laxity tests.13 Patients with isolated anterolateral rotatory instability were generally improved one grade on the Lachman scale while wearing the Lenox Hill knee orthosis. However, when there was combined instability, excessive anterolateral rotatory instability was unchanged. Hoffman, et al.,12 evaluated six commercially available knee orthoses for their ability to stabilize ligamentous injuries against valgus, axial rotation and the drawer sign symptoms using fresh cadaver specimens by measuring bony displacements as external loads were applied. Of the six knee orthoses evaluated by Hoffman, the 3D 3-Way knee orthosis (3D Orthopedic Inc., Dallas, TX) provided the greatest overall restriction of knee motion.
The American Academy of Orthopedic Surgeons22 reviewed knee orthoses organized into categories of prophylactic, rehabilitative and functional. Generally very little objective biomechanical or functional data regarding the orthoses were provided by the manufacturers. However, a cadaver study was reported wherein the resistance of four prophylactic orthoses to valgus loads was measured. They also reported that there was no limitation of the drawer sign by any of the four orthoses evaluated. In the category functional knee orthoses, Robert Hunter, M.D., compared the effectiveness of the Lenox Hill and the CTi knee orthoses in protection against the drawer sign in 15 subjects with an absent anterior cruciate ligament in one knee. The results showed that both orthoses held the knee to within normal limits at 15 pounds of anteriorly directed force at both 25 degrees and 90 degrees of knee flexion while neither provided protection at 20 pounds of force.
Because performance data are scarce and technical/clinical guidelines unavailable, it is difficult for a clinician to objectively select the appropriate knee orthosis for a given patient/pathology. Therefore, the purpose of this study is to objectively compare eight commonly prescribed knee orthoses for their ability to restrain excessive knee valgus, varus and axial rotation. A follow-up study will be directed at evaluating the drawer sign and recurvatum.
The specific aims of this study are to:
Develop an apparatus for evaluating valgus, varus and axial rotation;
Document the mean (+/- sd) of valgus, varus and axial rotation deformity for each of several torque loads for eight knee orthoses;
Compare the motion allowed among the eight knee orthoses at each of the torque loads;
Define the relationship between the torque and angle for each orthosis tested, in each position.
To objectively evaluate the eight knee orthoses, a unique testing apparatus was developed (Figure 1) . The apparatus consists of an above-knee prosthesis which had been modified to allow valgus, varus and axial rotation angles much greater than normal. The socket portion of the prosthesis was filled with polyester foam and a mounting pipe was inserted to hold the prosthesis horizontally. This pipe was inserted through two tee-joints, which were attached to a wood base. The prosthetic knee joint was removed and replaced with a custom-made joint constructed of soft crepe. This custom-made crepe knee joint required minimum torque to create valgus, varus and axial rotational deformities.
To simulate natural skin friction, a rubber sleeve was stretched over the prosthesis. This allowed the components and straps of the various knee orthoses to grip the otherwise smooth surface of the prosthesis.
With the test apparatus in the horizontal side-lying position, the tibial portion freely fell into 30 degrees of valgus or varus and 25 degrees of axial rotation (Figure 2 and Figure 3 ). Axial rotation was indicated by two offset alignment marks on the femoral and tibial sections of the prosthesis. The femoral portion of the AK prosthesis was immobilized and the tibial section was able to angulate medially or laterally independent of the femur when torque was applied. The instrumentation used consisted of a torque wrench (Model 1502DIN, Consolidated Devices Inc., City of Industry, CA), tensiometer (Model DDP-50, John Chatillon & Sons, Inc., Greensboro, NC), electrogoniometer (Neuromuscular Engineering Department, Rancho Los Amigos Medical Center, Downey, CA) and linear scale (Figure 4) . The electrogoniometer, which consisted of a truss assembly connected to a potentiometer and a digital readout, measured and indicated the angular deformities produced. Custom-made mounting brackets were used to secure the electrogoniometer for measuring angular deformity of the test prosthesis (Figure 5) . The brackets could be arranged to accommodate valgus, varus and axial rotation angulation measurements.
The eight knee orthoses selected for testing included the Polyaction, Lerman, Lenox Hill, ECKO, DonJoy, Analog, Pro Am and CTi (Table 1) . These orthoses were either purchased or donated. Although all existing knee orthoses were not evaluated, this group is considered to be representative of the knee orthoses presently available.
Preliminary calculations indicated that a 150 pound patient in single limb support with varying degrees of varus/valgus may require a knee orthosis to restrain approximately 500 inch-pounds of torque. Therefore, for the purpose of comparison, the applied valgus and varus torque was set to range from 0 to 650 inch-pounds, in ten 65 inch-pound increments. Similarly, the applied axial rotation torque ranged from 0 to 240 inch-pounds, in 60 inch-pound increments. These ranges of applied torques were selected to represent practical clinical conditions. The extreme torques associated with high level professional sports is unknown.
For each orthosis, the torque application sequence was repeated three times. This loading sequence was also done when no orthosis was on the testing apparatus. All testing took place with the test apparatus in the horizontal supine position.
Once a knee orthosis was placed on the test apparatus, axial rotation torque was applied with a torque wrench with an analog dial indicator, while valgus/varus torque was applied with a constant lever tensiometer. The lever arm for the valgus/varus tests was 13 inches from the point of application of the force to the knee joint axis, thus giving torque (inch-pounds) force (pounds) x 13 inches. Each angular deformity was digitally indicated on the electrogoniometer. This procedure was repeated three times for each deformity for each of the eight knee orthoses. Between trials, slippage and inelastic alignment were corrected.
For each value of valgus, varus or axial rotational torque applied, the angle that the fixture deformed, with or without an orthosis attached, was recorded.
Valgus and varus force was applied with the tensiometer at a point corresponding to the apex of the lateral malleolus (Figure 5) . As the fixture with the orthosis applied began to deform, the resulting angulation was displayed on the electrogoniometer digital readout. After each trial was performed, the electrogoniometer was reset to zero and the leg reset to its original position as indicated by alignment marks on the prosthesis and on the base board.
Axial rotational torque was applied with a torque wrench to a nut welded to the distal margin of the prosthesis (Figure 7) . The electrogoniometer mounting brackets were rotated 90 degrees from the previous location to record the angles produced by the deforming knee orthosis. Anterior alignment marks on the femur and tibia sections of the test fixture were used to reset the fixture to the starting point.
The data were screened for outliers and summarized. Since only one of each orthosis was tested, the three replicate measurements were averaged and that value used to represent the motion obtained for each orthosis at each position and torque load. Analysis of variance was used to test for variation between the orthoses. This was done separately for each motion at each of the imposed torques. Regression analyses were also used to detect significant linear trends between the variables of torque and angle. As the applied torques caused the apparatus and orthosis to deform, it was obvious that some of the resistance to deformity was caused by the apparatus. To isolate the resistance to motion caused by the orthoses alone, the resistance due to the apparatus was subtracted. This was done by subtracting the torque required by the apparatus alone at a specific angular deformity from the opposite torque required by the apparatus and orthosis being tested. The procedure was repeated at two degree increments from 0 to 50 degrees.
The statistical programming package Crunch (Crunch Software Corp., 5335 College Aye, Oakland, CA 94618) was used to perform the data analysis, and all testing was done at a .05 level of significance.
The mean (+/- sd) valgus angle of deformity ranged from a low of 0 degrees for the Polyaction, Lerman, ECKO, Analog and Pro Am orthoses at 65 inch pounds of torque to a high of 46.7 degrees (+/- 1.16) for the ECKO knee orthoses at 650 inch-pounds of torque (Table 2) . Post hoc testing using the Sheffe multiple comparisons test showed the Polyaction, Lerman and Analog knee orthoses had consistently greater ability to resist deformity, than the other five orthoses, through 390 inch-pounds of torque. At the load of 455 inch-pounds the Analog knee orthosis was joined by the CTi in showing the most significant resistance to deformity through the final load of 650 inch-pounds attaining a valgus angle of 13 degrees (+/- 1), 13 degrees (+/- 0) respectively. The Lennox Hill and ECKO knee orthoses showed the least resistance to deformity, where 6 degrees (+/- 1) and 7 degrees (+/- 1) of valgus motion was noted at 130 inch-pounds of applied torque. At 260 inch-pounds the DonJoy joined these two orthoses and continued to resist valgus motion significantly less than the other five knee orthoses through the final torque load of 650 inch-pounds, attaining 46.7 degrees (+/- 1.16), 42.7 degrees (+/- 1.16) and 38.7 degrees (+/-1.16) for the ECKO, Lenox Hill and DonJoy, respectively. The test prosthesis with no orthosis deformed 49 degrees (+/- 1) into valgus (maximum deformity permissible on the apparatus) at 325 inch-pounds and was significantly more flexible than all of the test orthoses (Table 2) .
The first load of 65 inch-pounds of torque was the only load at which all of the knee orthoses protected the knee to within the maximum comfortable valgus deformity of 3.4 degrees.4 The Polyaction provided the best protection by allowing less than 3.4 degrees of valgus through 260 inch-pounds of torque. The Polyaction, Lerman, Analog and CTi best restrained valgus to within 8 degrees, through 390 inch-pounds of torque, the angle which corresponds to internal knee disruption.3(Table 2)
Significant linear trends were detected for torque versus angle for each of the orthoses evaluated, (r .95 to .99). Further regression was applied and the best curve fit to these data were observed with third order regression (r .983 to .998); however, this was not significant (Figure 7) .
The Polyaction, Lerman, Analog and Pro Am knee orthoses showed significantly greater resistance to varus deformity from 65 to 260 inch-pounds of applied torque; 7 degrees (+/- 0), 9.3 degrees (+/- 2.08), 8.3 degrees (+/- .58), and 9.7 degrees (+/- 1.16), respectively, at 260 inch-pounds (Table 3) . At 325 inch-pounds the Polyaction yielded 11.3 degrees (+/- 1.53) and the Analog yielded 9 degrees (+/- 0) and both were significantly more resistant to varus motion than any of the other orthoses. From 390 to 650 inch-pounds of applied torque the Analog was, alone, significantly more resistant to varus deformity than the other seven orthoses, attaining a maximum varus deformity of 15.7 degrees (+/- .58). The ECKO knee orthosis exhibited the greatest degree of flexibility being significantly more deformed, 49.7 degrees (+/- .58), at the maximum torque load. The ECKO, DonJoy, Lenox Hill, and CTi deformed significantly more than all of the other orthoses for varus loads of 195 inch-pounds to 325 inch-pounds. The test prosthesis, with no orthosis, deformed 50.7 degrees (+/- 1.53) into varus (maximum deformity permissible on the apparatus), at a load of 455 inch-pounds, which was significantly greater than any of the test orthoses.
At the maximum comfortable varus angle, 3.4 degrees,4 the Polyaction provided the best protection (195 inch-pounds) while the ECKO, Lenox Hill and DonJoy provided the best protection by exceeding the 3.4 degrees maximum comfortable varus range in degrees after only 65 inch-pounds of applied load. At the minimum angle at which internal joint disruption occurs (8 degrees3), four orthoses - Polyaction, Lerman, Analog and Pro Am - were stiffer than the others holding through 260 inch-pounds of varus torque. The ECKO provided the least protection by exceeding the 8 degrees after the first torque load.
Again a significant linear trend was observed between the variables of torque and angle for each of the orthoses (r .979 to .99). With third order regression applied, there was an improvement in curve fit (r .991 to .999) which, again, was not significant (Figure 8).
There are two groups of applied rotational torques: 0 to 60 inch-pounds, to evaluate the test apparatus with no orthosis; and 30 to 240 inch-pounds used to evaluate the eight knee orthoses.
The greatest variation between the eight knee orthoses occurred in axial rotation. The only axial rotation load at which all eight orthoses could be compared was 60 inchpounds (Table 4) . At this axial rotation load, the Polyaction and CTi deformed 2.3 degrees (+/-1.16) and 2.3 degrees (+/-.58), respectively; both of these knee orthoses were significantly more resistant to axial rotation motion than any of the others. The ECKO and DonJoy orthoses allowed significantly more motion, 40.3 degrees (+/- 1.53) and 38.3 degrees (+/- 1.16), respectively, at the first load of 60 inch-pounds, and exceeded the maximum permissible axial rotational deformity at the second rotational torque of 120 inch-pounds. Of the remaining six knee orthoses, the Lennox Hill was significantly more flexible than any of the others, allowing 30.3 degrees (+/-3.51) of axial rotation at 120 inchpounds. At 180 and 240 inch-pounds of applied torque, the CTi and Lerman were significantly more resistant to axial motion, attaining 31 degrees (+/- 1) and 34.7 degrees (+/- 1.53), respectively, at the maximum load.
The maximum comfortable axial rotation angle of 25 degrees5 was maintained up to 180 inch-pounds by only the Lerman and CTi knee orthoses. The axial rotation angle at which internal knee disruption occurs (37 degrees) was maintained up to 240 inch-pounds of applied axial rotation torque by only the Lerman and CTi knee orthoses.3
There was significant linear relationship between the applied torque and resulting angle (r .93 to .99) for all of the orthoses tested for rotation, which improved as third order regression was applied (r = .996 to .999). This change was also, not significant.
The Analog and CTi knee orthoses exhibited the overall best resistance to high valgus loads (Table 2) . It is interesting to note that although the Analog and CTi knee orthoses offer the greatest resistance to valgus at high loads; the Polyaction, Lerman and Analog were better able to avoid the 3.4 degrees maximum comfortable deformity up to 195 inch-pounds of torque (Table 2 and Figure 7 ).
The Analog knee orthosis exhibited the overall best resistance to varus torques (Table 3 and Figure 8 ), especially at the higher varus loads (>325 inch-pounds). As was the case for valgus, the Polyaction knee orthosis was best able to avoid the maximum comfortable varus deformity (3.4 degrees) up to applied loads of 195 inch-pounds.
In general there was a 25 percent greater resistance to valgus than varus; the reason for this is not clear.
The CTi and Lerman knee orthoses exhibited the overall best resistance to axial rotation torques (Table 4 and Figure 9 ), including the axial angles corresponding to the maximum comfortable rotation (25 degrees) and protected completely against the axial angle at which internal joint disruption occurs (37 degrees). Regression curves of the axial rotation data (Figure 9) show that the CTi and Lerman knee orthoses were 25 to 30 percent better than the other six knee orthoses tested. This study did not support the finding of Knutzen, et al.21 which showed that the Lenox Hill restricted axial rotation of the knee to approximately 7 degrees, during weight-bearing while walking. While these two studies are not comparable, the Lenox Hill exceeded 7 degrees of axial rotation at the first torque load which is calculated to beless than that experienced during the weight-bearing phase of gait.
Only two of the knee orthoses tested required a plaster impression to be sent to the manufacturer: the Analog and Lenox Hill. The CTi knee orthosis required an outline of the leg be drawn by a company representative. Therefore, these three were considered custom made and all the others were "off the shelf." In spite of being custom fabricated, the Lenox Hill knee orthosis did not fit well, which may account for its poor overall performance. Conversely, the Pro Am knee orthosis was not custom fabricated, did not fit well and performed surprisingly.
In general the more rigid the knee orthosis, the more resistant it was to deforming forces. Rigidity depended upon design (CTi), the use of metal sidebars (Analog, Polyaction and Lerman), and overall length (leverage). If the knee orthosis fits well, then good resistance deformity at small loads was more likely and corrective pressures were less (not measured).
The best knee orthosis in terms of fit and cosmesis, the ECKO, did not perform the best. It remains an obvious problem that the cosmetically "conscious" knee orthosis may not necessarily be functionally capable of doing the job for which it was intended; yet the functionally capable knee orthosis may not be acceptable to the patient or prescriber.
It is the authors' opinion that before a given knee orthosis can be objectively recommended for a given patient/pathology, standards should be devised as to how much knee angle deformity can be tolerated and the magnitude of external loads which must be restrained. It is obvious from Teitz, et al.19 that the orthoses used by the football players did not provide adequate protection. However, to determine if protection by a knee orthosis is even possible, the specific loads imposed during the activity of football need to be known. It may be that the most restrictive knee orthoses may not be capable of guarding against high deforming torques. Therefore, the best knee orthosis may not be good enough. However, it may be found that the least restrictive knee orthoses may suffice for a wide range of low torque applications.
Thomas R. Lunsford is Chief Orthotist at Rancho Los Amigos Medical Center, Downey, California, and Clinical Director of the Orthotics and Prosthetics Baccalaureate Program at California State University, Dominguez Hills.
Brenda Rae Lunsford is Biostatistician and Manager of Information Systems, Physical Therapy Department, Rancho Los Amigos Medical Center, Downey, California.
Jack Greenfield is Assistant Chief of the Orthotics Department at Rancho Los Amigos Medical Center, Downey, California.
Sharron E. Ross, currently a resident orthotist at Rancho Los Amigos Medical Center, Downey, California, is a graduate of the University of California, Riverside, and also a graduate of the California State University, Dominguez Hills, where this study was conducted in conjunction with the Orthotics and Prosthetics Baccalaureate Program requirements.