Evaluation of the gravity-correction feature of a Kin-Com isokinetic dynamometer.

[Finucane SDG, Mayhew TP, Rothstein JM. Evaluation of the gravity correction feature of a Kin-Come[R] isokinetic dynamometer Phys Ther. 1994,74:1125--1133.!

Key Words: Equipment, exercise; Muscle performance, lower extremity; Tests and measurements, functional

Measurements of forces are frequently obtained with electromechanical dynamometers. These measurements are sometimes reported as percentages. Ratios or percentages are often used to compare measurements obtained from the same muscle of opposite limbs (eg, uninvolved/involved)[1] or measurements obtained from agonist and antagonist muscles groups (eg, hamstring/quadriceps femoris).[2] Ratios such as these have been shown to be error-ridden when the effects of gravity on the limb are not taken into account. Methods for correcting these errors have been suggested.[3-5] The weight of the limb segment that produces a moment (rotational force) must be added to the force (or torque) measurements when the muscle is attempting to move the limb segment against gravity and subtracted when the muscle is attempting to move the limb segment in the direction of the gravitational forces. We will use the term "gravity-correction factor" to mean the amount of force (or torque) that is added to or subtracted from the measured force (or torque). The Kin-Com[R] and other similar devices measure the component of forces that produce moments about their lever arm axes. The magnitude of the rotational component of gravitational forces changes with the angle of the lever arm (Fig. 1).

Many electromechanical dynamometers in use today are controlled by computers and include proprietary software. Many of the software programs include procedures for correcting measurements for the effects of gravity. Often the algorithms used in the gravity-correction routines are not provided by the manufacturer, and data on the accuracy and reliability of the measurements obtained with these procedures are lacking.

The importance of using gravity-correction procedures in certain circumstances is generally recognized. To our knowledge, no studies have been performed that have determined the most appropriate method of weighing a limb segment. Manufacturers' guidelines are generally based on theory and practicality rather than data. A variety of methods of weighing the leg prior to measuring force (or torque) produced by the thigh muscles have been described in the literature.[2,6-8] Some of these methods weigh the limb while it is in motion by allowing the leg to be passively dropped through a specified range of motion[6] or by having an active robotic device move the leg through a specified range of motion.[7] Other authors[2] have described placing the knee at a specified angle other than full extension in order to weigh the limb.

Many of the gravity-correction procedures are poorly described. For example, Rathon et al,[8] using a Kin-Com[R] dynamometer,(*) weighed the limb with the knee in full extension and with the lever arm horizontal. When the leg is attached to the force pad of the Kin-Com[R] lever arm, the leg and the lever arm are not parallel (Fig. 2). Therefore, we have difficulty understanding how both conditions described by Rathon et al could be met simultaneously. Often investigators state that they followed the manufacturer's guidelines when weighing their subjects' leg[9] or simply state that gravity-correction procedures were done.[10]

The gravity-correction procedure for the Kin-Com[R] dynamometer is described in the user's guide[11] and in the proprietary software program (version 3.21) through menu selections on the computer monitor. The procedures described in these two sources, however, are different in regard to the position of the limb segment when determining the gravity-correction factor. The Kin-Com[R] user's guide instructs the user to "move the lever arm (with the patient's limb attached) as close to horizontal as possible with no muscular tension."[11] The procedure outlined by the computer software directs the user to "move the limb as close to the horizontal position as possible with no muscular tension." Because the limb is not parallel to the lever arm, these two positions are not equivalent (Fig. 2). A discrepancy, therefore, exists between the two procedures suggested by the manufacturer of this device. Because the positions of the limb segment would be different in the two procedures, the gravity-correction factors that are obtained may be different. This discrepancy points to the need for testing of these procedures for each device, and also to the inadequacy of simply stating that "manufacturer's guidelines were followed." Other versions of the Kin-Com[R] II user's guide and software may not contain the descriptions described in this article. Users of all dynamometers should determine whether the directions provided for correcting for gravitational forces are appropriate.

Both gravity-correction procedures described by the manufacturer include the caution that there should be "no muscular tension." Presumably, passive forces across a muscle/tendon unit act on the Kin-Com[R] strain gauge, creating error in measurements of limb weight. To our knowledge, the effects of these passive forces on gravity correction have not been described in the literature. in our laboratory we have dealt with this perceived source of error by positioning subjects supine rather than sitting during measurements of lower-limb weight. Theoretically, there should be less passive tension across the hamstring muscle/tendon unit with the hip extended than with the hip flexed. This issue must be studied as part of the process of identifying the best positions for weighing limbs during gravity-correction procedures.

Preliminary testing of a Kin-Com[R] dynamometer (model 500-11 with software version 3.01) in our laboratory identified sources of error in the measurements of the rotational component of gravitational forces that had not been previously described. The general purpose of this study was to distinguish between errors of measurements of gravitational forces inherent to the Kin-Com[R] and errors in the measurement of gravity-correction factors that are unique to measurements of human limbs. Three questions were addressed relating to measurements obtained with the gravity-correction procedure of the Kin-Com[R] dynamometer: (1) What was the effect of lever arm position on measurements of the rotational component of gravitational forces acting on a known weight applied to the lever arm? (2) What was the effect of lever arm position on measurements of the rotational component of gravitational forces acting on a limb segment? and (3) What was the effect of trunk position on measurements of the rotational component of gravitational forces acting on a limb segment? Thus, our study was performed in two palls. The purpose of part 1 of the study was to determine the effects of different Kin-Com[R] lever arm positions on force measurements. The purpose of part 2 of our study was to examine the effects of different Kin-Com[R] lever arm positions and subject trunk positions on measurements of the rotational component of gravitational forces acting on a relaxed limb segment.

Methods

Instrumentation

A Kin-Com[R] dynamometer (model 500-11 with software version 3.21) was used in all portions of this study. The Kin-Com[R] is built such that strain gauges are aligned to measure forces that are in a direction perpendicular to unit's lever arm. All angular measurements will, therefore, refer to the angle of the lever arm from horizontal unless otherwise specified. In part 1 of our study, the pad was removed from the lever arm attachment so that a weight pan could be suspended from the lever arm attachment (Fig. 3). The center of the attachment was marked to standardize the position of the weight pan.

Part 1

The gravity-correction procedure of the Kin-Com[R] evaluation protocol was used to determine the effect of lever arm angle on measurements of force of a known weight. The evaluation protocol of the Kin-Com[R] software was followed. This protocol includes an option for collecting data for the gravity-correction procedure. The joint setting on the Kin-Com[R] device's menu was set for the knee, the movement pattern setting for extension was selected, the lever arm length was 35 cm, and the gravity-correction option was turned on. The overlay procedure of the isokinetic protocol (part of the evaluation procedure) was used. The lever arm was placed in the horizontal position using a gravity-referenced protractor,([dagger]) and the procedure for anatomical reference was completed by following the menu instructions. The position with the lever arm horizontal was identified as 0 degrees (Fig. 3), and angular positions of the lever arm below the horizontal (toward the floor) were identified as positive.

The next step in the Kin-Com[R] evaluation menu was the gravity-correction procedure. We moved the lever arm to the position for testing and locked it into place by pressing the enter key on the keyboard. A weight pan was hung from the lever arm attachment, and a 44-N weight was placed on the weight pan. The weight was allowed to stabilize, and the force of the weight (including the weight of the weight pan and lever arm attachment) as measured by the Kin-Com[R] strain gauge (MF) was read from the computer monitor and recorded. The force as measured by the Kin-Com[R] was recorded with the lever arm positioned at 5-degree intervals from 0 to 50 degrees. After recording the weight at a given angle, the weight was removed and the lever arm was moved to the next position. The procedure was repeated on a second day.

The weight of 44 N was chosen because this was a convenient single weight that fell within the, range previously observed for limb weights. In a previous series of studies in our laboratory, limb weights (including the lever arm attachment) for 28 subjects ranged from 37 to 76 N (unpublished data). All weights within a reasonable range should act the same on the Kin-Com[R] strain gauges. A single weight, therefore, was considered adequate for this study, which was designed to examine the technical performance of a device. The reliability of measurements of weights has been established.[12] The intraclass correlation coefficient (ICC) (version 3,1) for reliability of measured force between days for weights ranging from 22 to 450 N was .99. The coefficient of determination ([r.sup.2]) for the strength of the linear relationship between recorded weight and actual weight was .99.[12]

Part 2

A 23-year-old male subject (weight=86 kg, height=183 cm) was used for this portion of the study. The subject sat on the Kin-Com[R] bench in the position to test his right knee musculature. The backrest was positioned so that the subject's hips were at approximately 70 degrees of flexion. The hip angle was determined by visual approximation. His right thigh was strapped to the stabilization pad. The Kin-Com[R] was adjusted so that the subject's lateral femoral epicondyle was aligned with the center of rotation of the device's lever arm. The shin pad of the Kin-Com[R] was attached to the subject's leg approximately 2 cm above the medial malleolus, which resulted in a 32-cm lever arm.

The Kin-Com[R] evaluation procedure was again used to test the gravity-correction procedure (see "Part 1" section). The limb segment was weighed at 12 positions: with the knee in full extension (the lever arm was above the horizontal) (Fig. 2B) and with the lever arm positioned in 5-degree increments from 0 to 50 degrees. The procedure was then repeated with the backrest removed and the subject positioned supine. The angle of knee flexion was measured with a standard goniometer at the 0-degree lever arm position.

Data Analysis

Part 1. The between-days reliability of the force measured by the Kin-Com[R] dynamometer (MF) was determined by comparing the MF for the 2 days using the ICC (3,J).[13] The ICC (3,1) was used to assess the error associated with our Kin-Com[R].

The magnitude of MF should have been dependent on the angle of the lever arm (Fig. 1). We calculated the rotational component of the force at each angle for each day based on the measured force at 0 degrees, using the following equation:

(1) PF = (MF at O[degree]) x cos [theta]

where PF was the predicted rotational component of the gravitational force, [theta] was the angle of the lever arm, and MF at 0 degrees was the measured force when the lever arm was horizontal. The MF at 0 degrees was used as the actual weight because in this position the weight pan was perpendicular to the lever arm and the entire force of gravity acted to rotate the lever arm (Fig. 1). The reliability of MF at 0 degrees for freely hanging weights has been established.12 The PF was the force that should have been measured by the Kin-Com[R] at each lever arm position. The relationship of PF to MF was described by using the method of least squares to determine the linear regression coefficient for each day. A t statistic was used to test whether the slopes of the lines of best fit were different from 1 and whether the intercepts were different from 0.[14] The 0-degree position was not included in the analysis because by definition PF at 0 degrees was equal to MF at 0 degrees.

The percentage of error (% error) associated with measurements of force at different angles was calculated as follows:

(2) % error = (PF-MF)/PF X 100

The percentage of error was plotted against angle in order to examine the relationship between the changing lever arm angle and the error in measurement. Linear regression coefficients were determined by the methods of least squares to describe this relationship for each day. A t statistic was used to determine whether the slopes or intercepts for these lines were different from 0.

Part 2. The data for this portion of the study were examined in a similar manner. The predicted rotational component of the gravitational force (PF) was calculated for each angle using equation 1. The MF at 0 degrees (lever arm horizontal, Fig. 2A) was used as the actual weight of the limb segment. The relationship of PF to MF was described, determining linear regression coefficients by the method of least squares for each trunk position. We determined whether the slopes of these lines were different from 1 and whether the intercepts were different from 0 using a t statistic.[14]

The percentage of error for the relationship of MF to PF for each angle was calculated using equation 2. Linear regression coefficients were again determined to describe this relationship. The t statistic was used to determine whether the slopes and intercepts for the lines of best fit were different than 0.

The lever arm positions at 0 degrees and with the knee extended were not included in the linear regression analysis. By definition, MF and PF at 0 degrees were equal. Based on our observations of the plot of percentage of error to lever arm angle, the error for the knee-extended position (lever arm above the horizontal) clearly represented a different phenomenon than the error when the lever arm was below the horizontal.

The MF at 0 degrees in sitting and supine positions were compared. This comparison allowed us to make some inferences about the effect of placing the hamstring muscles at two different lengths. The MF with the knee extended was compared with the MF at 0 degrees. This comparison was done because both of these positions are recommended by the manufacturer of the Kin-Com[R]; however, there was a difference in the lever arm angle with these two positions. Examination of this relationship also allowed us to look at the effects of fully extending the knee on the measurements of force due to gravity on the leg.

Results

Part 1

The MF recorded for each of the 2 days, the PF based on the MF at 0 degrees for each day, and the percentage of error at each angle for each day are displayed in Table 1. The ICC (3,1) for MF on day 1 compared with day 2 was .90. The linear regression equations for these relationships were MF = 0.97 x PF - 2.51 N ([r.sup.2] =.96) for day 1 and MF 0.83 x PF + 9.33 N ([r.sup.2]=.93) for day 2. If force measurements by the Kin-Com[R] (MF) were free from error at varying angular positions (ie, MF equal to PF), the slopes for the regression lines would have been 1.00 and the intercepts would have been 0 N. The slopes were not significantly different from 1 for day 1 (t= -0.354, P>.05) or day 2 (t=2.139, P>.05). The intercepts were not significantly different from 0 for day 1 (t= -0.734, P>.05) or day 2 (t=0.0302, P>.05).

[TABULAR DATA 1 OMITTED]

The relationship between angle of measurement and percentage of error for both days is depicted in Figure 4. The linear regression equation describing this relationship was % error = 0.05 x [theta] 6.76 ([r.sup. 2] =.04) and error = - 0.12 x [theta] + 1.00 ([r.sup.2] =.39) for days 1 and 2, respectively. The slopes were not significantly different from 0 for day 1 (t=0.546, P>.05) or for day 2 (t= -2.277, P>.05). The intercepts for these lines were also not significantly different from 0 (t=0.023 and t=0.724 for days 1 and 2, respectively; P>.05). The low coefficients of determination ([r.sup.2]) and the fact that the slopes of these lines were not different than 0 indicate that percentage of error was not related to angle of measurement. With the exception of measures of force at 45 degrees on day 1 and 50 degrees on day 2, all MF values were within 10% of the predicted force at the respective angular position.

Part 2

The MF, PF, and percentage of error for each lever arm angle for the sitting and supine positions are presented in Tables 2 and 3. The PF and percentage of error for the lever arm angle of 0 degrees are not given because all other PFs were calculated based on the MF at 0 degrees.

Table 2. Part 2 of Study: Measured and Predicted Forces and Percentages of Error for the Subject's Leg Suspended From the Shin Pad of the Kin-Com[R] With Subject in Sitting Position

[[theta](a)        MF(b)    PF(c)    % Error(d)
-23[degrees](e)     94       55.2       -70.2
  0[degrees]        60       ...        ...
  5[degrees]        58       59.8         3.0
 10[degrees]        54       59.1         8.6
 15[degrees]        46       58.0        20.6
 20[degrees]        47       56.4        16.6
 25[degrees]        40       54.4        26.4
 30[degrees]        32       52.0        38.4
 35[degrees]        33       49.2        32.9
 40[degrees]        28       46.0        39.1
 45[degrees]        22       42.4        48.2
 50[degrees]        19       38.6        50.7

(a) [theta]=lever arm angle (in degrees).

(b) Mf=measured rotational component of gravitational force (in newtons).

(c) Pf=predicted rotational component of gravitational force (in newtons).

(d) % error=error of the measured force based on predicted force for each angle.

(e) Angle of lever arm, above the horizontal, with subject's limb at full extension.

Table 3. Part 2 of Study.. Measured and Predicted Forces and Percentages of Error for the Subject's Leg Suspended From the Shin Pad of the Kin-Com[R] With Subject, in Supine Position

[theta](a)        MF(b)   PF(c)   %Error(d)
-17 [degrees](e)   80     52.6      -52.1
  0 [degrees]      55     ...        ...
  5 [degrees]      49     54.8       10.6
 10 [degrees]      42     54.1       22.5
 15 [degrees]      42     53.1       20.9
 20 [degrees]      35     51.7       32.3
 25 [degrees]      32     49.9       35.8
 30 [degrees]      31     47.6       34.9
 35 [degrees]      26     45.1       42.3
 40 [degrees]      23     42.1       45.4
 45 [degrees]      15     38.9       61.4
 50 [degrees]      11     35.4       68.9

(a) [theta]=lever arm angle (in degrees).

(b) MF=measured rotational component of gravitational force (in newtons).

(c) PF=predicted rotational component of gravitational force (in newtons).

(d) % error=error of the measured force based on predicted force for each angle.

(e) Angle of lever arm, above the horizontal, with subject's limb at full extension.

The linear regression equations for the relationship of MF to PF were 1.74 x PF - 51.64 N ([r.sup.2] =.93) for the sitting position and 1.76 x PF - 52.83 N ([r.sup.2]=.96) for the supine position. Again, these relationships do not include the knee extended position or the position with the lever arm at 0 degrees. The slopes for each line were significantly different from 1 and the intercepts were different from 0 for each of the trunk positions. For the sitting position, t=4.25 (P<.05) for slope and t=-5.69 (P<.05) for the intercept. For the supine position, t=0.636 (P<.05) for the slope and t=-8.55 (P<.05) for the intercept.

The relationship between percentage of error and angle of measurement is depicted in Figure 4. The linear regression equations for this relationship were % error = 1.04 x [theta]- 0.02 ([r.sup.2]=.95) for the sitting position and % error = 1.15 x [theta] + 5.86 ([r.sup.2]-.94) for the supine position, with positions of 0 degrees and knee extension omitted. The slopes for both of these lines were different than 0 (t=11.962 for the sitting position and t= 11.111 for the supine position; P<.05). The intercepts were not significantly different than 0 for either of these lines (t= - 0.075 for the sitting position and t= 1.832 for the supine position; P>.05). Thus, as the angle of measurement increased, the percentage of error also increased. The percentage of error was greater than 10% for all angles greater than 10 degrees with the subject sitting and for all angles other than 0 degrees with the subject in the supine position. For all lever arm positions below the horizontal, the measured force was less than the predicted force for each angle.

When the knee was in full extension, the lever arm was 17 degrees above the horizontal for the supine position and 23 degrees above the horizontal for the sitting position. In each position, the MF was greater with the knee extended than with the lever arm at 0 degrees. This increase was from 55 to 80 N in the supine position and from 60 to 94 N in the sitting position. The PF, given these lever arm positions, was 55.5 N for the sitting position and 52.6 N for the supine position for the knee-extended position. There was a small difference in the MF at 0 degrees in the sitting position compared with the supine position. The MF at 0 degrees was 55 N in the supine position and 60 N in the sitting position.

Discussion

Using the gravity-correction procedure, there was good agreement between the force measured by the Kin-Com[R] and the predicted rotational component of the gravitational forces when a weight was suspended from the lever arm. This agreement was apparent for lever arm angles from 0 degrees (horizontal) to 50 degrees below the horizontal. There was no agreement between MF and PF at all angles when the same procedure was used to measure the force due to gravity of a subject's limb attached to the lever arm. The linear regressions and coefficients of determination describing the relationship between percentage of error and lever arm angle indicate that when the weight was loaded onto the lever arm, there was no relationship between the lever arm angle and the percentage of error. Thus, the error did not increase with increasing lever arm angles. There was, however, a strong relationship between percentage of error and lever arm angle for measurements of the subject's limb. The slope of the linear regression line was greater than 0 for both subject trunk positions. Thus, the amount of error increased with increasing knee flexion when the dynamometer was loaded with a subject's limb. Errors of greater than 50% were observed with lever arm angles of 45 to 50 degrees.

The gravity-correction procedure recommends placing the limb segment

or the lever arm in the horizontal position to determine the weight of the limb segment (MF at 0 [degrees]). The software apparently uses a cosine formula to predict the rotational component of the gravitational forces (gravity-correction factor) for each angular position. This gravity-correction factor is added to or subtracted from the torque measurements at each angle to determine the total amount of muscular torque produced during a movement or isometric contraction. Investigators do not always weigh the limb with the lever arm in the horizontal position.[2,6,7] When the limb segment is weighed at an angle other than 0 degrees, the Kin-Com[R] software still calculates the rotational component of the gravitational force using a cosine function, the measured force, and the angle at which the lever arm was set during the measurement.

With the subject's limb attached to the lever arm attachment, there was a progressively larger error in the measurement of the rotational component of force as the limb was moved from a position with the lever arm at 0 degrees (horizontal) to 50 degrees below the horizontal. The difference between the PF and MF was present regardless of whether the subject was positioned sitting or supine. Although we did not attempt to determine why the percentage of error increased as the knee flexion angle increased, there appear to be three possibilities. First, as knee flexion increased, passive stretch of the quadriceps femoris muscle may have placed an upward force on the limb, resulting in the MF being lower than the PF. This effect would likely be greatest with the hips extended (ie, with the subject positioned supine). Second, if there was an alteration in the alignment between the leg and the lever arm as the knee flexed, this may have affected the MF. Finally, the leg cannot hang in the same manner as a weight pan from the lever arm. Unlike the weight pan that is attached only to the lever arm, the leg is also attached to the knee. The gravity-correction procedure is based on the assumption that the limb hangs like a pendulum, orienting the weight in line with the gravitational vector, so that the full predicted rotational component of the gravitational force would act perpendicular to the lever arm. Because the limb is not suspended as a pendulum from the lever arm, some of the force of the limb will act on the knee rather than on the lever arm or will act along the length of the lever arm and not at the strain gauge. In either case, this decreases the amount of the force acting perpendicular to the lever arm, resulting in the measured force being less than the predicted force.

We do not know whether other electromechanical dynamometers would also demonstrate this error. We would recommend that users of these other devices investigate the potential for error when gravity correction is calculated at an angle below or above the horizontal. Error may also be present in protocols in which the gravity-correction procedures require the limb to be passively dropped or moved through a range of motion.

The results of our study provide information regarding the effects of hamstring muscle length on forces measured during gravity-correction procedures. The subject in this study was able to obtain only a 70-degree straight leg raise on the right side. Thus, is knee extension or hip flexion increased, the passive tension in the hamstring muscle/tendon unit would have increased. The effects of this increasing tension are seen both in the comparison of MF with the lever arm horizontal and the knee extended and in the comparison of MF with the subject in the sitting and supine positions.

The measured force of the limb segment with the knee fully extended, for both trunk positions, was greater than with the lever arm horizontal and the knee slightly flexed. Based on the formula for determining gravity-correction factors, the MF with the knee fully extended (lever arm above the horizontal) should have been less than the MF with the lever arm horizontal (Tabs. 2 and 3, Fig. 1). The strain gauge, attached to the lever arm, measures forces perpendicular to the lever arm, rather than forces that are perpendicular to the limb. Therefore, gravity-corrected data would be flawed if the knee-mended position were used with a subject with tight hamstring muscles.

There was a small difference in the force recorded with the lever arm at horizontal when the subject was sitting versus when he was positioned supine. This difference may have again been due to tightness of the hamstring muscles. The sitting position, with the knee extended and hip flexed, stretched the hamstring muscles across two joints. The amount of knee flexion (approximately 20 [degrees]) and the amount of hip flexion (approximately 70 [degrees]) in the sitting position may have been sufficient to create some passive tension across the hamstring muscles. The supine position would have relieved this stretch. Similar results might be expected whenever a muscle is placed in a position of stretch.

The instructions supplied with the Kin-Com[R] II and III dynamometers are inconsistent with respect to the procedure for gravity correction. The directions in the Kin-Com[R] user's manual instruct the user to move the lever arm as dose to the horizontal as possible without inducing muscular tension.[11] The computer software directions instruct the user to move the subject's limb segment as close to the horizontal as possible without inducing muscular tension. The results of this study demonstrate that these two positions are not equivalent.

Conclusions

Based on the results of this study, we would recommend that users of the Kin-Com[R] dynamometer perform the gravity-correction procedure for the leg with the lever arm near the horizontal position, and not the limb segment at the horizontal position knee extension). When there is any concern that the subject or patient might have short hamstring muscles, the individual should be placed in a supine position rather than the traditional sitting position. We would recommend that users of other systems replicate this study with their system to determine whether the same problems are apparent when knee flexion increases. Finally, we recommend that the joint position or lever arm position used during gravity-correction procedures be reported in order to make studies or day-to-day uses of the Kin-Com[R] replicable.

Although this was a bench test, and did not include multiple trials with multiple subjects, we believe that the results are important because the use of electromechanical dynamometers in physical therapy clinics is common. Therapists using these devices for creating percentage comparisons between different muscle groups of the same limb, the same muscle group of two limbs, or muscles of different individuals should be using gravity-correction procedures.[3-5] The procedures used must provide the most accurate estimation of limb weight in order to give the best correction of force measurements for the effects of gravity. The errors reported here have not been previously reported and, therefore, needed to be brought to the attention to users of electromechanical devices.

Acknowledgments

We thank M Scott Sullivan for his assistance with the statistical analysis.

(*) Chattercx Corp, Part of Chattanooga Group Inc, 4717 Hixson Rd, PO Box 489, Hixson, TN 34343.

[dagger] Sears Craftsman Protractor, Sears, Roebuck & Co, Sears Tower, Chicago, IL 60684.

[FIGURES 1 to 4 ILLUSTRATION OMITTED]

References

[1] Kannus P, Jarvinen M, Johnson R, et al. Function of the quadriceps and hamstrings muscles in knees with chronic partial deficiency of the anterior cruciate ligament. Am J Sports Med 1992;20:162-168. [2] Worrell TV, Perrin DH, Gansneder BM, Gieck JH. Comparison of isokinetic strength and flexibility measures between hamstring injured and non-injured athletes. J Orthop Sports Phys Ther. 1991;13:118-125. [3] Winter DA, Wells RP, Orr GW. Errors in the use of isokinetic dynamometers. Eur J Appl Physiol. 1981;46:397-408. [4] Nelson SG, Duncan PW. Correction of isokinetic and isometric torque recordings for the effects of gravity: a clinical report. Phys Ther. 1983;63:674-676. [5] Fillyaw M, Bevins T, Fernandez L. Importance of correcting isokinetic peak torque for the effect of gravity when calculating knee flexor to extensor muscle ratios. Phys Ther. 1986;66:23-29. [6] Gleason NP, Mercer THE. Reproducibility of isokinetic leg strength and endurance characteristics of adult men and women. Eur J Appl Physiol. 1992;65:221-228. [7] Durand A, Malouin F, Richards CL, Bravo G. Intertrial reliability of work measurements recorded during concentric isokinetic knee extension and flexion in subjects with and without meniscal tears. Phys Ther. 1991;71:804-812. [8] Rathon JA, Matthews KM, Yang AN, et al. Effects of different acceleration and deceleration rates on isokinetic performance of the knee extensors. J Orthop Sports Phys Ther 1991;14: 161-168. [9] Peterseri SP, Bell GJ, Bagnall KM, Quinney HA. The effects of concentric resistance training on eccentric peak torque and muscle cross-sectional area. J Ortbop Sports Phys Ther. 1991;13:132-137. [10] Overend TJ, Cunningham DA, Kramer JF, et al. Knee extensor and knee flexor strength: cross-sectional area ratios in young and elderly men. J Gerontol. 1992;47:M2O4-M210. [11] Kin-Com[R] II and III Muscle Testing and Training System: Clinical Desk Reference. Hixson, Tenn: Chattecx Corp, Part of Chattanooga Group Inc; 1992. [12] Mayhew TP, Rothstein JM, Finucane SDG, Lamb RL. Performance characteristics of the Kin-Com[R] dynamometer. Phys Ther. 1994;74: 1047-1054. [13] Shrout PE, Fleiss L. Intraclass correlations: uses in assessing rater reliability, Psychol Bull. 1979;86:420-428. [14] Neter J, Wasserman W, Kutner MH. Applied Linear Statistical Models Regression, Analysis of Variance, and Experimental Designs. 3rd ed. Homewood, III: Richard D Irwin Inc; 1990: 62-72.

SDG Finucane, PhD, PT, is Assistant Professor, Department of Physical Therapy, Medical College of Virginia, Virginia Commonwealth University, Box 980224, Richmond, VA 23298-0224 (USA). Address all correspondence to Dr Finucane.

TP Mayhew, PhD. PT, is Assistant Professor, Department of Physical Therapy, Medical College of Virginia, Virginia Commonwealth University.

JM Rothstein, PhD, PT, FAPTA, is Professor and Head, Department of Physical Therapy, College of Associated Heath Professions, University of Illinois at Chicago, 1919 W Taylor St, Chicago, IL 60612. He is also Professor of Biomedical Engineering, University of Illinois at Chicago, and Chief of Physical Therapy Services, University of Illinois Hospital.

This article was submitted by July 30, 1993, and was accepted June 23, 1994.