The influence of subject and test design on dynamometric measurements of extremity muscles.

The importance and widespread use of dynamometry is evidenced by the large number of references to dynamometric measurements in the physical therapy literature. Since 1988, 30 to 40 publications per annum have reported findings based on data from electromechanical dynamometers. Electromechanical dynamometers have been used for many purposes, although data supporting these uses have not always been provided. Among the uses were: 1. To collect normative values for muscles and for various types of subjects.[1-14] 2. To classify muscle performance as normal or abnormal by comparisons with the performance of contralateral muscles,[15-18] with normative data,[19,20] or with muscle performance in a control group.[20-22] 3. To collect torque curves that might indicate whether pathology or characteristics specific to subject type were present.[19,21-30] 4. To establish the relative efficacy of various treatment and training regimens.[23,24,31-44] 5. To quantify exercise so that exercise regimens may be administered.[23,35,38,39,41,45-50] 6. To evaluate the effects of training or testing modes (eg, eccentric,concentric, isometric),[33,34 42,46,51-55] testing or training speed,[46,48,52,57-61] and duration of training [24,39] 7. To investigate factors that correlate with measurements. Measurements have been compared with muscle cross-sectional area measured by computerized tomography,[48,50,62-64] associated electromyographic activity,[38,65-72] type or location of electrical stimulation that causes force production,[73,75] physiological factors associated with muscle performance,[47,76-88] and biomechanical factors associated with muscle performance.[63,82,89,90] 8. To investigate the relationship between dynamometric measurements and measurements obtained with other tests.[23,24,31,34,42,44,82,91-93

Electromechanical dynamometers are also in extensive clinical use.[94,95] Clinicians involved in the assessment or treatment of persons with disabilities use such dynamometers to determine the need for intervention, the extent of impairment, and changes in subject performance. Dynamometric measurements have also been used in courts of law as evidence of functional capacity.[94] Muscle performance throughout the tested range of movement can be examined with some dynamometers, whereas this was not possible with previously available instruments. Force produced at constant lever arm speed (isokinetic testing) and the speed achieved when resistance to movement is held constant (isotonic testing) can also be measured.

Factors That Influence Measurements

Factors that may affect measurements need to be identified, and these effects need to be appreciated by clinicians and researchers. Changes in measurements should reflect changes in a subject's ability. If the procedures used influence measurements, test procedures need to be replicated when testing occurs on multiple occasions. Comparisons with published data are appropriate only when similar procedures are used. Similarly, when researchers obtain measurements under different conditions, comparison of results should be possible only if these differences have an inconsequential effect on the magnitude of the measurements.

In an early review on this subject, Mayhew and Rothstein[96] argued that measurements are influenced by several aspects of test procedure, including method of axis alignment, damp settings, and whether the measurements are corrected for the effects of gravity. They emphasized that the conditions under which measurements are generated must be defined if the measurements are to be meaningfully interpreted. Seventy-five percent of the publications we examined, however, were published after 1985, the majority of which were published after 1988.

A recent review of this topic[97] examined the literature on subject stabilization and test position, contraction mode, choice of equipment, and method of axis alignment. Our search of the literature, however, indicated a wider body of literature than has been considered in previous reviews, which provides evidence on a broader range of variables that could affect test outcome. We examined articles that presented information on measurement changes resulting from alteration of a single aspect of the testing circumstances. A subject or test factor was considered capable of influencing measurements when research findings indicated that systematic differences among mean values occurred when that factor alone was systematically altered. The reports reviewed were located using both manual and electronic searches of the literature. Electronic searches of Index Medicus were conducted primarily using combinations of the key words "dynamometer," "dynamometry," "strength," "muscle," "tests," "testing," "isokinetic," "normative," and "data."

Subject-Related Factors

Age

Dynamometric measurements of young and older subjects have been compared,[2,7,98-101] and measurements have also been reported for subjects in adjacent age decades.[5,13,102] These independent investigations all reported that forces decline with increasing age. This was true despite subject sample and methodological differences.

Details of test procedures to study the relationships between age and measurements were rarely found to be complete, a finding common to the broader body of literature reporting dynamometry testing. Omissions included descriptions of the rest intervals separating tests, how lever arm length was determined, how axes were aligned,[5,7,60,102] whether gravity correction was performed[2,5,7,60,102] the type of warm-up used,[2,7,60,102,103] whether signal damping occurred,[7,60, 102,103] whether feedback was given,[5,103] and the type of equipment used.[102]

Incomplete protocol description prohibits study replication. When test procedures cannot be precisely replicated, comparisons of reported measurements with measurements for subsequent tests of individuals may be misleading. In addition, the magnitude of age-related differences could be altered by differences in test design. At present, however, little evidence supports the possibility that experimental conclusions regarding the influence of subject age on measurements would differ if studies were replicated under conditions that varied from the conditions in the original studies. Presumably, the experimental conditions were standardized for each subject tested within a single investigation. It is improbable, therefore, that systematic alterations in test procedures caused the systematic changes in measurements associated with subject age consistently reported in these independent studies.

The reliability of obtained measurements was sometimes not reported in studies of aged muscle.[13,60,102] When reliability of measurements was reported, a variety of approaches were used. Some authors[2,5,101] preferred to use correlation coefficients, although it was not always clear whether the obtained correlation was for between-or within-session measurements.[5] Other approaches included reporting differences between measurements obtained in repeated tests as a percentage of those measurements[2,103] or reporting means and standard deviations of differences between scores obtained within and between sessions.[7] A discussion of these various options for presenting the error associated with measurements is beyond the scope of this review. Failure to ascertain the magnitude of error associated with measurements, however, does not necessarily negate conclusions based on data. The studies cited determined whether differences existed between measurements obtained for subjects grouped according to age. Findings of no significant differences, a statistical conclusion, have two possible explanations. One explanation is that measurements for the groups under comparison were similar. The other explanation is that although the groups were different, the magnitude of random error variability associated with measurements obscured systematic differences between groups of differing ages. The possibility would then exist that a more sensitive measurement, one with greater reliability, might detect such differences. When significant differences were detected between groups, these systematic effects were apparent despite random variability in measurements, however great. Regardless of the magnitude of random error associated with the obtained measurements, the magnitude of systematic effects attributable, in this case, to the age of subjects, was apparent. The need to establish reliability is important for many reasons.[104] In cases where statistically significant differences between means exist, however, the failure to first establish the magnitude of error associated with the measurements does not detract from the conclusions that group differences were found.

There are, however, some competing explanations for force declines associated with increased subject age. Because of the difficulty associated with longitudinal studies of the effects of aging on force, almost all investigations have compared groups of younger subjects with groups of older subjects. In order for differences between groups to be attributed to age, the groups need to be similar for other relevant criteria. One such variable is the weight of subjects tested.[3,60,105] At least for subjects under 33 years of age, weight has been positively correlated with torque production.[60,105,106] Differences in subject weight are unlikely to account for the effects of age observed by several researchers[2,5,101] because torque declined, even though the older subjects, on average, were heavier than the younger subjects. Although total body weight[5,101] may not reflect muscle mass, the same age-related torque declines were observed when subjects in different groups were matched for lean body weight[2,100] or when measurements were corrected for fat-free mass.[107]

A perhaps more viable explanation for age-related declines in measurements is that the activity levels of subjects of different ages also vary and therefore what is being observed relates to declining activity levels. Laforest et all[100] found that measurements of knee extension and knee flexion for both young and older tennis players were greater than measurements obtained for sedentary subjects who were matched for age, height, and lean body weight. Young tennis players, however, had greater force production compared with older tennis players, and young sedentary subjects had greater force production compared with older sedentary subjects. These results reinforce the likelyhood that age and activity level may combine to influence measurements, confounding the relative influence of each. The majority of studies on age-related changes in torque did not provide any indication of subject activity level.[2,5,7,13] Research that clearly differentiates the relative contribution of activity and aging to force measurements presents a substantial but important challenge.

Another explanation for the observed deterioration of torque with advancing age is that older subjects may require different test procedures in order to produce maximum values. Frontera et al[98] found that subjects aged from 45 to 78 years produced greater torques during a second test session than they produced during the first test session. Because only older subjects were tested, it cannot be determined whether the increases were unique to older subjects or generalizable to other test conditions, but the possibility remains that practice requirements for performance of the test may differ for subjects of differing ages. Similarly, it has not been established whether optimum warm-up and rest intervals differ for subjects of different ages.

Decreases in measurements of torque associated with advancing age have been observed. Whether age alone accounts for this deterioration or whether other factors that commonly occur in conjunction with aging are the cause is not clear. Investigations into age-related muscle changes provide some explanations for the observed force declines, including age-related decreases in the ability of muscle to generate tension,[99-101,108] decreases in muscle mass[99,101,109-111] decreases in proportion and selective atrophy of type II fibers,[107] and increases in nonmuscle tissue in muscle compartments.[101]

The magnitude of age-related changes, being specific to the sample tested, may have little application to tests of different individuals. In addition, the magnitude of age-related changes has been reported to vary with test conditions such as test position[102]; muscle group tested[2,103]; and whether the test was isometric or isokinetic,[7] or concentric or eccentric.[2] Despite differences in procedure, several researchers[7,100,101,107-112] have obtained force measurements for subjects who were in their seventh decade of life that were 70% to 80% of measurements obtained for subjects who were in their third decade of life Rice et al[113] found an annual decline in measurements that averaged 2% across subjects aged 62 to 102 years. Although information regarding the reliability of the unorthodox measurement procedures used was not provided, the results were supported by other investigations.[114,115] Because many subjects were in nursing homes, the rate of decline probably reflects changes that occur in part as a function of having a relatively inactive lifestyle.

That significant differences in measurements have been observed when the same test procedures were applied to groups of subjects of different ages suggests that subject age should be considered when test results are compared. Clinicians and researchers comparing results for tests of individuals or groups with those reported in the literature should be aware that age differences among subjects under comparison can confound interpretation of measurement differences. Comparison of test results with those reported in the literature would be enhanced by knowledge of the mean, standard deviation, and range of ages of the subjects tested. Reports that detail only the range of ages of subjects[7,13,116-113] or that omit the range of age[6,9,10,119-121] are commonly published Some reports of "normative data" omitted any information regarding the age of subjects.[4,11,122,123] Additionally the majority of published age-specific "normative data" is for groups spanning at least 10 years of age.[5,7,13,117,124] For those decades when strength gains or losses might be expected, research that attempts to narrow down the years when these changes more commonly occur may be useful. Additionally, interactions have been reported between age-related changes and other aspects of test procedures. Thus, for age-related normative data to be used for comparison with the results of tests of individuals, details regarding the test procedure used in the collection of such data must be available to clinicians.

Weight

Body weight has been reported to have an influence on the magnitude of dynamometric measurements,[3,60,105] with heavier subjects tending to produce higher values than lighter subjects.[60,105,106] Although the existence of such a relationship has intuitive appeal, several factors should guide conclusions based on this body of literature. In none of the investigations cited were values corrected for the influence of gravity. For movements with gravity, measurements obtained for heavier subjects are likely to be artificially inflated because of the greater weight of their body segment. In addition, omissions in descriptions of protocols used were common and included the rest intervals between tests,[3,60] the warm-up procedures used,[60] the method used to align axes,[3,60,105] and whether the measurements were reliable.[3,60]

Furthermore, the method for determination of subject weight varied. Some researchers[3,60,105] although not specifying how subject weight was determined, presumably used total body weight. Other researchers[106] used a variety of weight estimates, including fat-free mass. Hoffman et al[125] correlated lean body weight with dynamometric measurements instead of using total body weight because they believed that total weight (ie, including fat) should correlate less well with subject strength. This argument may be logical, but the use of estimates of fat-free mass may be of less consequence when lean subjects are tested. Hortobagyi et al[106] found that correlations between weight and torque measurements for athletic subjects were similar, regardless of whether total body weight or fat-free mass was used. In contrast, results were found to be influenced by the method used to determine weight when "somewhat obese" children were tested.[126] Estimating lean body weight is more labor intensive than determining total body weight, and further research on the value of normalizing force measurements to lean body weight in preference to total body weight would appear to be beneficial. If total weight is reported, however, it may be appropriate to provide some evidence that subjects tested are not atypical with respect to percentage of body fat. In addition, torques expressed as a percentage of any measure of body weight should only be compared with measurements that were similarly calculated.

Subject weight has repeatedly been correlated to measurements of knee flexion and extension, and a range of coefficients that represent the relationship between the two factors have been reported.[60,105,106] As Hortobagyi et al[106] pointed out, the reported correlation coefficients must be interpreted cautiously. As the range of subject weights increases, the effect of the relationship between weight and torque measurements is likely to increase. If very heavy and very light subjects are tested, subject weight is likely to be a better predictor of performance than it would be for subjects of similar weight. The correlations obtained between weight and torque measurements for pooled male and female data,[105] for subjects spanning a large age range,3 or for a mixture of trained and untrained subjects, therefore, may not represent the correlations expected when weight and torque measurements are correlated for subjects who are more homogeneous with respect to weight. Hortobagyi et al[106] found nonsignificant correlations between weight and dynamometric measurements of knee flexion and extension for a group of athletes who were relatively homogeneous with respect to age and training as well as weight. They concluded that factors other than muscle size may play at equally important roles in explaining differences in muscular strength.

Subject weight accounts for some of the variability in torque measurements, and this variability increases as the range of subjects' weights increases.[106] Muscle cross-sectional area partially accounts for variability in force measurements, and weight to some extent reflects muscle mass.[101] Researchers intending to provide useful data should report the weight of subjects. The range and distribution of force measurements obtained by researchers for subjects of similar weight would be of additional value to clinicians who want to compare the results of tests of individuals with published data.

Interaction Between Gender and Weight

Investigations[1,2,125,127,128] indicate that torque measurements obtained with a dynamometer for male subjects exceed those obtained using the same protocol to test female subjects of similar age and athletic background. Comparison of torque measurements, therefore, should be made with consideration of gender.

It has been suggested that some gender-related differences can be eliminated by expressing measurements as a function of weight.[105,125] Hoffman et al[125] obtained torque measurements for bench press and leg press per unit of lean body weight for 30 male and 30 female cadets. Only the mean age of subjects was reported (male subjects, X=20.2 years; female subjects, X=19.4 years). Much of the detail regarding the test protocol was not reported. Additionally, the reliability of the measurements was not reported. The error associated with the measurements may have concealed gender-related differences in weight-adjusted leg-press values. Regardless of the magnitude of error associated with the test procedures, however, the systematic differences in weight-adjusted values attributable to gender were of sufficient magnitude to attain significance for bench press measurements. The authors concluded that when measurements are expressed per unit of body weight, leg strength differences between male and female subjects were within random variation, but the arm torques of men were higher. The authors argued that gender-related differences are not solely a function of overall subject size. Hoffman et al[125] proposed that circumference and bone-diameter measurements determine lean body weight.[129] These measurements are similar for men and women in the hip region but differ substantially in the shoulder region, suggesting that proportions of lean body weight by body region vary with gender.

Falkel[3] reported that weight and not gender was a critical factor in determining torque for isokinetic tests of planter flexion. The results cannot be extrapolated to represent an absence of gender-related differences in mature adults, however, because two thirds of the subjects tested by Falkel were under 16 years of age. In addition, as the error associated with the obtained measurements was not determined, the failure to attain significance for gender effects could represent insensitivity of the measurements. Furthermore, replication of the Falkel test could be difficult as the authors failed to clarify how axis alignment was determined, whether gravity correction was applied to the measurements, and what rest intervals separated contractions.

Nicholas et al[117] found that "peak torque per body weight" for knee extension (tested at 60[degrees]/s), hip extension (at 30[degrees]/s), and hip flexion (at 30[degrees]/s) did not differ with gender. Differences were found, however, for knee extension (at 180[degrees]/s), knee flexion (at 60[degrees] and 180[degrees]/s), and shoulder extension and shoulder flexion (at 60[degrees] and 180[degrees]/s). Some reservation regarding findings of nonsignificant differences is warranted, given the small sample size tested and the fact that several subjects failed to complete some of the test movements. In addition, no indication of the measurement error associated with the test procedure was provided. If score variability was inflated in both groups by random variability, this may account for the failure of some of the observed differences to attain statistical significance. Nevertheless, however large the measurement error, the measurements were sufficiently sensitive to reveal a systematic effect of gender for weight-adjusted values for some of the muscles tested. This finding indicates that for some lower-limb muscles, gender differences may not be erased by weight adjustment of measurements and there is the possibility that gender differences in weight-adjusted measurements may not be uniform across different movements or test procedures. Again, the application of the results is limited by omissions in the protocol description regarding the rest intervals allowed between contractions; whether data were corrected for the effects of gravity; and how subject stabilization, axis alignment, and lever arm length were determined.

Given the conflicting research findings, gender-related differences in test results should be considered likely for all tests and possible for any weight-adjusted measurements. It therefore appears inappropriate to combine male and female subjects' weight-adjusted measurements prior to analysis because "when weight is statistically controlled for, there is no difference in lower-extremity strength between sexes."[105](p91)

When interpreting the results of studies, caution should be exercised when statistical analyses reveal that no differences exist between groups. This finding may indicate that there is no difference between groups, or it may reflect statistical insensitivity in identifying an effect. Such insensitivity is influenced by the sample size, the effect size, and the magnitude of random variability in measurements. We believe that an indication of this problem exists when we see conflicting findings from authors.[5,102] Highenboten et al[5] found concentric peak torque/body weight ratios for knee extension tests at 50[degrees]/s in subjects aged 25 to 34 years to be greater for male subjects (X=2.49) than for female subjects (X=1.98). Fischer et al,[102] who calculated peak torque/ body weight ratios for isometric knee extension tests in subjects aged 20 to 29 years (X=2.8 and 2.1 for male and female subjects, respectively), found that forces produced by male subjects did not differ from forces produced by female subjects. The differences between mean values for male and female subjects appear to be similar in both experiments. The conflicting conclusions may be due to differences in measurement sensitivity, experimental design, or the statistics used to examine the data.

In summary, forces generated by male subjects generally exceed those generated by female subjects when male and female subjects are matched for age and athletic background. Thus, when measurements are reported in research reports, combining data from male and female subjects[91,130-133] may limit the value of the results. Some authors have reported that weight-adjusted lower-limb measurements from male and female subjects do not differ, whereas other authors have found differences for specific tests. We believe that conclusions cannot be generalized beyond the test conditions reported. We contend that when weight-adjusted measurements from male and female subjects are pooled for analysis, interpretation of results should include the consideration that gender differences may exist.

Athletic Background of Subjects

The influence of participation in athletics on force measurements has been investigated in several ways. Measurements obtained when the same test conditions were applied to athletes and matched control subjects have been compared. Alternatively, force measurements of subjects pursuing different athletic endeavors have been examined. In addition, measurements obtained under particular test conditions have been compared with measurements obtained under different test circumstances of subjects with different athletic backgrounds. The results indicate that participation in athletics influences force production. There are several factors, however, that influence the conclusiveness of many of these studies.

Although investigators who have compared measurements from athletes with measurements from matched control subjects or other types of athletes have usually compared subjects of similar ages, they have often not matched subjects with respect to weight.[134,137] Conclusions regarding the effect of participation in athletics on measurements have sometimes been based on tests of so few subjects[134,137] that the generalizability of findings must be made with caution. Many authors failed to report the reliability of measurements.[11,117,135-137] Although such an omission does not negate findings of group differences, it does limit confidence that test results would be similar if the investigation were repeated. Again, not a single report was sufficiently detailed to enable precise replication. Among the omissions were how axis alignment was determined

[11,100,117,135-137] whether measurements were corrected for the effect of gravity,[11,117,135] the rest intervals allowed between repetitions,[11,117] and how subjects were stabilized.[117]

Nicholas et al[117] compared test results for a small group of nonathletic subjects with the results of several earlier studies of athletic and nonathletic subjects. The table of mean values obtained for tests of different samples demonstrates the considerable overlap in measurements for athletes and untrained subjects. In addition, there appears to be marked variability among measurements, even for samples of presumably similar subjects. Sample differences may account for the range of reported mean measurements obtained in these independent investigations. Another possible explanation is that experimental procedures differed in each investigation. Examples of such procedural differences include the amount of rest allowed between tests, the type of warm-up permitted, the number of test repetitions used, and the type of stabilization used. Most reports failed to adequately define the test procedure used; thus, the reason for observed differences in measurements cannot be confidently concluded.

Alexander[1] tested concentric knee extension of elite sprinters and concluded that the sample tested produced torques that were substantially greater than those of the nonathletic subjects tested by Francis and Hoobler.[138] Although this conclusion may be true for the studies compared, the generalizability of these findings (de, sprinters produce greater forces than do nonathletic subjects) is limited due to important differences in the two groups under comparison. These differences were in subject age, scale used for measurement, and test speed. Because age[2,5,7,13,102] and concentric test speeds have been shown to influence results, the differences in measurements between groups solely attributable to subject differences cannot be isolated. Therefore, although it may be true that sprinters produce greater forces than do nonathletic subjects, this has not been conclusively demonstrated and we certainly have little idea about the magnitude of this difference.

Although conclusions based on dynamometry tests of athletes is somewhat confounded by a lack of information regarding the test procedures, many investigations have in common the finding that athletes generate greater torques than do nonathletes. Although this finding may appear obvious, overlap of measurements for athletic and nonathletic populations occurs. If systematic differences between groups are to be attributed to the influence of athletic participation, the circumstances under which measurements are collected must be similar. This reinforces the need for researchers to clearly and adequately describe subject and test details if dynamometry is to provide definitive information regarding effects of this nature.

Height of Subjects

Molnar and Alexander[146] found a positive correlation between knee and elbow flexor and extensor torques and height in children aged 7 to 15 years. This correlation was reported to be more consistent and potent than the correlations to age, weight, sitting height, biacromial diameter, or calf circumference. Tabin et al,[126] however, in tests of 10-to 15-year-old children, found height to be only "loosely" correlated with measurements, with stronger correlations between measurements and lean body weight. Again, differences in experimental procedure make the conflicting conclusions difficult to evaluate. These differences were in sample size, subject age, muscles tested, and possibly method used to determine body weight. Moreover, replication of either of these experiments may not be possible due to the limited description of test procedures. Neither research group reported the test range of movement, subject position for testing, type of warm-up, rest intervals between test repetitions, preload or damping used, or whether measurements were gravity corrected. Thus, the relationship between height of children and measurements for specific test conditions remains uncertain.

Presence of Impairment

The influence of impairments on dynamometric measurements has been investigated by comparing measurements for an injured limb with those for a contralateral healthy limb.[15,18,43,44,63,147-149] These studies have found measurements for injured limbs to be lower than those for noninjured limbs. The results of these investigations therefore support the notion that dynamometric measurements can reflect anticipated weakness in an injured limb. There are, however, several factors that limit the usefulness of this research. Half of the reports inadequately documented the test procedures used.[18,44,63,148] It is impossible, therefore, to assess the technical competence of the research or to validate the results through replication or through application of experimental procedures. No reports included estimates of the magnitude of error associated with measurements. Although such an omission does not invalidate findings of lower forces for the impaired limb, confidence that research findings would be replicated if the investigation was repeated on another occasion would be enhanced by evidence of the temporal stability of measurements.

These results suggest that measurements obtained with a dynamometer are sensitive enough to reflect strength differences between limbs in individuals with unilateral injuries. These results do not, however, provide clinicians with information regarding the amount of difference between measurements for contralateral limbs that constitutes evidence of impairment. The claim has been made that differences between measurements for contralateral limbs of greater than 10% provide evidence of impairment or muscle imbalance,[18,25,63] but the evidence to support such a claim does not appear to be strong. Mira et al[18] made such a claim based on knee extension tests of 15 volunteers without any known impairment. A quadriceps femoris muscle strength index was determined for each limb for each subject. This index was derived by pooling an average of isokinetic and isometric measurements, although the mathematics used were not reported. Strength indexes between contralateral limbs differed by a mean of 6%, with a standard deviation of 4%. These results led the authors to conclude that a mean variation of more than 10% represents abnormal differences between limbs. The generalizability of the findings are limited by the following factors. There was no description of the protocol used. Not even the test speed used was reported. The type of measurement was not described. No argument was made for the use of pooled isokinetic and isometric values, and the sample was too small to be considered representative of the population. One standard deviation of values around the group mean was used to derive the 10% figure. If 95% confidence in the application of these findings is desirable, the estimate of between-limb differences based on this work would more appropriately be considered 14%.

Goslin and Charteris[4] tested the knee extension of 30 male and 30 female subjects. The dominant limb was defined as the limb producing the greater knee extension torques. Nondominant-limb knee extension torques were then reported as a percentage of dominant-limb torques. The mean percentage difference was 86.7% (SD=14.4%) for the male subjects and 81.2% (SD=10.7%) for the female subjects. The results of this study are difficult to interpret. In the absence of evidence of reliability of the data, it is impossible to argue that the source of the observed difference between two limbs was not measurement error. Most likely, the observed effect was a combination of both limb strength differences and measurement error. The generalizability of the findings is also questionable. Protocol description was limited, and no demographic information on subjects other than gender was provided. The results suggest, however, that measurements obtained on a single occasion can, in subjects without known impairments, differ between limbs by more than 10%.

Grace and colleagues' research[25] further supports the inadequacy of the 10% guideline. These researchers tested the knee extension and flexion of 172 highschool football players aged 15 to 17 years. Ipsilateral-contralateral imbalance was defined as a difference in the isokinetic measurements between sides of 10% or more, as calculated by dividing the difference between the side of the greatest magnitude and the side of lesser magnitude by the side of the greatest magnitude and multiplying the result by 100. Grace et al reported that imbalances of more than 10% were a frequent finding. Interestingly, no relationship was identified between percentage differences between measurements for opposite limbs and subsequent injury during the football season. Interpretation of this report is again confounded by the inadequate description of protocols and a lack of evidence for the reliability of the measurements.

Differences that can be expected when testing subjects without impairments need to be reported. Such reports need to include evidence of measurement stability, the conditions under which data are collected, the measurements used for comparison between limbs, and how percentage differences are calculated. A broad range of subjects, with adequate representation of different genders, ages, weights, and athletic backgrounds, might also assist the subsequent application of results to tests of individuals.

Expressing the lower value as a percentage of the higher value or the right-limb measurement as a percentage of the left-limb measurement may or may not be an appropriate approach to the provision of clinically useful information. Use of percentage differences between limbs is based on the assumption that stronger subjects without impairments can be expected to demonstrate greater absolute strength differences between contralateral limbs than weaker subjects without impairments and that the differences are meaningful. For example, if it is claimed that contralateral peak torque differences for knee extension tests of subjects without impairments are within 10%, a subject producing 200 N.m of torque during a right knee extension test would be considered unimpaired if he or she produced 180 N.m of torque during a left knee extension test. A subject producing 50 N.m of torque during a right leg test would have to produce 45 N.m of torque during a left leg test to be considered unimpaired. No investigations were found that examined the relationship between the magnitude of a subject's measurement and the expected differences between measurements for contralateral limbs. Another approach might be to determine the absolute magnitude of differences between measurements for contralateral limbs demonstrated by subjects without impairments.

The influence of impairment on measurements obtained with a dynamometer has also been investigated by comparing measurements from injured subjects with measurements from matched control (noninjured) subjects.[10,20,21,64,150] There are many difficulties associated with interpretation of the results of these studies. No reports included estimations of the temporal instability of measurements. Components of protocol design essential to experiment replication were frequently not reported. Omissions included whether data were corrected for the effect of gravity,[10] how axis alignment was determined,[21] and how subjects were stabilized.[21] In some reports, no description of protocols was provided.[20-64] In addition, whether the lower values obtained from the impaired subjects were different from those from the control subjects was not always clear.[20,64] Although most researchers selected control subjects whose age and weight were similar to those of the impaired subjects, or reported weight-adjusted data, it was not always clear how well control subjects and impaired subjects were matched for athletic ability.[10,21,64] Instructions given to subjects may also influence obtained measurements. Dvir et al,[21] for example, reported that persons with patellofemoral pain were instructed not to push through pain. The results therefore provide a comparison between the pain-free capability of persons with patellofemoral pain and the maximum capability of noninjured control subjects. In the other studies cited, instructions given to injured subjects were not routinely clarified.

Attempts to compare measurements from impaired subjects with those from unimpaired control subjects raises the issue of the usefulness of normative data. Even for subjects who are homogeneous with respect to age, weight, and athletic background, large differences in subject capability can be anticipated. Refinement in our understanding of the methods under which normative data can be collected that minimize the differences in measurements currently observed between subjects appears to be a prerequisite for useful comparative data.

Uncertainty still exists regarding the magnitude of difference between measurements for opposite limbs that constitutes evidence of impairment. Even more limited are attempts to define impairment and disability by comparison of measurements with normative data from nonimpaired subjects. Investigations into the possibilities for dynamometry as a diagnostic tool have been hampered by inadequate reporting of research procedures and failure to establish the reliability of measurements. Collecting usefully large amounts of data on injured subjects may prove a daunting task for any single researcher. With adequate documentation, however, research can be replicated, validated, and extended by other investigators. The opportunity to collate results of multiple investigations has the potential to yield a larger body of data that better represents measurement differences attributable to specific types of disability and to provide clinicians with guidelines for evaluating tests of individuals. Such an approach to construction of a useful body of data requires that test procedures and subject details be reported in sufficient detail to enable accurate replication of test conditions.

Limb Dominance

Investigations into the influence of lower-limb dominance on muscle force measurements have used a variety of tests to determine limb dominance. Determination of lower-limb dominance has been made on the basis of hand dominance,[14,151] the limb used to kick a ball,[46,52,58,152] the limb used to kick a ball through two goalposts set 1.2 m (4 ft) apart,[142] the limb used to kick a ball through a doorway from a distance of 3 m (10 ft),[153] the stronger limb as determined by dynamometric testing,[4] the "preferred leg,"[121] or without specification of the criteria.[21,25,52,103,117,154] Lucca and Kline[155] attempted to develop a more elaborate test for limb dominance. Subjects were asked to perform five tasks, and the lower limb used to initiate each task was recorded. These activities were ascending stairs, descending stairs, kicking a ball, hopping, and picking up a pencil with the toes. The limb that initiated four out of the five tasks was considered the dominant limb. The protocol appears to be flawed by the crediting of dominance to the limb that initiated climbing as well as descending stairs, as the stronger limb might lead up but not necessarily down stairs. Unfortunately, preference as determined by this protocol was not correlated with other methods of identifying lower-limb dominance. This test identified the limb that was preferred for initiating an activity as the dominant limb. Another way to determine lower-limb dominance might be to identify the limb that is preferentially selected for demanding or enduring tasks. Further research into the comparability of the variety of procedures currently used for identifying lower-limb dominance appears to be warranted.

Investigations into the influence of lower-limb dominance on force measurements have dealt primarily with muscles that surround the knee joint.([double dagger]) In general, results indicate that no differences are apparent when measurements for the dominant lower limb are compared with those for the nondominant lower limb.[sections] Measurements for muscles other than knee flexors and extensors may provide additional insight into possible strength differences between dominant and nondominant lower limbs.

Limb dominance may influence the magnitude of upper-limb measurements. Hinton[16] found peak torque for shoulder medial (internal) rotation to be greater on the dominant side than on the nondominant side for pitchers. Shoulder extension has also been associated with greater measurements on the dominant side in pitchers, swimmers, and nonathletes.[124] In neither of these studies were data corrected for the effect of gravity. Without gravity correction, results could be confounded if the dominant arm was systematically heavier than the nondominant arm. It is unlikely, however, that the contribution by gravity accounts for the observed effects, as Hinton[16] found apparently similar differences between limbs when subjects were tested in a standing position and medial rotation was performed across gravity. Although neither Hinton nor Perrin et al provided evidence that measurements were reliable, at least on the single test occasion reported, systematic effects attributable to dominance were apparent regardless of the magnitude of error associated with the measurements. The method used to determine the axis of the shoulder movement was not stated in either report. Whether the alignment method was the same for each subject and not systematically affected by factors associated with the dominant limb cannot be determined from the reports.

The magnitude of dynamometric measurements, therefore, appears to be influenced by a number of subject factors. Clinicians or researchers wishing to compare measurements obtained from individual subjects with some criterion measurement should observe the age, gender, weight, athletic background, impairment, and possibly limb dominance of the subjects under comparison.

Movement-Related Factors

Joint Angle

Torques will vary depending on the joint angle at which data are collected.[18,156] Methods used to determine joint angle are neither standardized nor consistently reported. For example, in tests of muscles surrounding the knee joint, zero knee extension has been defined as maximal active knee extension,[157] full passive knee extension,[158] and full extension (active or passive not defined).[21,130,159] Sometimes, how joint angle is determined is not specified.[131,132,160] Wilhite et al[161] used goniometric measurements to designate 5- and 90-degree knee flexion, but they did not specify the structures with which the goniometer was aligned. It may be reasonable to propose that full passive and full active knee extension could lead to establishment of a tibiofemoral relationship at zero knee extension that differs by as much as 15 degrees. Verification of this proposal nevertheless awaits the definitive experiment. It would appear appropriate in the interim to at least report the method used in the determination of joint angle in order to facilitate replication of this aspect of the experimental method.

Muscle Action (Concentric, Eccentric)

Isokinetic testing can be performed with concentric or eccentric contractions. Measurements for maximum concentric contractions of specified muscles differ from measurements of eccentric contractions,[23,26,162] with eccentric peak torque being greater than concentric peak torque.[parallel] Effects observed in measurements obtained concentrically should not be assumed for eccentric test results. In addition, as test speed increases, concentric measurements tend to decrease and eccentric measurements tend not to change or tend to increase.[128] The ratio of concentric to eccentric measurements, therefore, will vary depending on test speed.[1,147]

Vyse and Kramer[56] investigated the effect of muscle action sequence on the magnitude of obtained measurements. They compared measurements obtained when elbow flexors were tested concentrically-eccentrically or eccentrically-concentrically. Eccentric tests that followed concentric tests resulted in peak torque measurements that were significantly lower (about 10%) than those produced using the eccentric-concentric sequence.

Mode (Isokinetic, Isometric, Isotonic)

Dynamometers can be used to test muscles isokinetically, isometrically, or isotonically. These different test modes affect the magnitude and type of test measurements obtained[3,57,87,144,165,166] Furthermore, the magnitude of the difference between isometric and isokinetic measurements has been reported to vary with test speed[143] and with the muscle group tested.[165]

Despite the knowledge that measurements vary with test mode, no research-based guidelines for clinical selection of test mode appear to exist. When clinicians test an individual, they must make a decision regarding the test mode to use. Evidence of impairment or change in performance indicated under the chosen test mode cannot be assumed for tests using optional modes. Until more information is available, clinical and research conclusions should not be extrapolated beyond the test mode used.

Pretesting Procedures

Warm-up Procedures

Most descriptions of test protocols include descriptions of whether subjects performed any kind of warm-up prior to testing, although exceptions to this are common.(#) Various kinds of warm-up procedures are reported, but the reasons for the use of one method of warm-up rather than another are not necessarily stated. Methods used for warm-up include an unspecified number of submaximal contractions[1]; several submaximal contractions[3,23,174]; between 2 and 4 submaximal contractions(**); 5 submaximal contractions[5,46,130,134,147,177]; 10 submaximal contractions36; and a series of submaximal contractions (4,[157] 6,[178] or 10[159]), which successively approach maximal effort. Some authors have chosen a combination of submaximal and maximal contractions for warm-up (2 submaximal, 1 maximal[131]; 3 submaximal, 1 maximal[49]; 3-4 submaximal, 1 maximal[77,132]; 3 submaximal, 2 maximal[153,179]; 3 submaximal, 3 maximal[58,180]; 5 submaximal, 1 maximal[51]; 5 submaximal, 2 maximal[94]; 5 submaximal, 5 maximal[105]; 5-8 submaximal, then practice trials[5]; 8-12 submaximal, 2 maximal[181]).

Some authors have used more subjective approaches to the warm-up protocol. Examples of this include allowing subjects to repeat submaximal contractions until they felt ready to perform maximally,[1] until the subjects demonstrated an understanding of the procedure by exhibiting smooth torque curves,[80] or when the reproducibility in repeated tests was deemed by the researchers to be good.[37] Some authors[5,16,22,52,128,178,179] have used exercises or workouts on bicycle ergometers prior to dynamometric testing. Wilhite et al[161] had subjects perform three repetitions each of a modified hurdler's stretch for hamstring muscle flexibility and a prone stretch for quadriceps femoris muscle flexibility prior to testing these muscles. Prior to testing at a range of speeds, Montgomery et al[182] positioned the subject for testing and the dynamometer was set for continuous passive motion at 120% / s for 5 minutes.

Presumably, the primary aims of warm-up procedures are to ensure subject safety during testing and to facilitate stability of test data. If optimum warm-up requirements to achieve these goals exist, they clearly have not been identified. Few investigations have specifically addressed optimum warm-up requirements. Almost 2 decades ago, Johnson and Siegel[116] tested knee extension using three submaximal warm-up contractions and six maximal test repetitions. The first three maximal contractions showed an increasing linear trend, followed by stable data. The authors suggested, therefore, that under the test conditions reported, three submaximal and three maximal repetitions were required to establish data without systematic trends.

Similarly, Mawdsley and Croft[183] found that maximal efforts tended to increase following three submaximal warm-up contractions. When subjects had no submaximal warm-up trials, however, the first three trials appeared to vary less and the second trial was the highest. Mawdsley and Croft[183] therefore argued that measurements for the average of the first three maximal trials may be reduced following submaximal warm-up repetitions. They proposed that submaximal warm-ups may not provide the most appropriate type of practice for their test conditions. They also noted, however, that some subjects who did not perform submaximal trials prior to maximal trials experienced knee discomfort. This was not a complaint when submaximal warm-ups were used.

Kues et al[184] obtained knee extension torques for 10 female subjects on three separate occasions. Based on visual analysis of data, they concluded that two practice sessions facilitate maximum performances. Although preliminary in nature, their report provides guidelines for researchers who want to determine the stability of their data prior to experimental intervention. Despite evidence that suggests the need for determination of the warm-up requirements for the specific set of test conditions used, researchers generally fail to provide justification for the selected warm-up procedures or evidence that obtained data is without systematic trends.

Subject Starting Position

Ample evidence has been provided that torque production is influenced by the position in which subjects are tested.([dagger]) ([dagger]) Although many investigators[102,128,130,145,185,186] failed to determine the temporal stability of measurements, the weight of evidence supports the likelihood that, when other test factors are held constant, changes to subject position alone can influence force measurements. The most probable explanation for this is that changes to the length of muscle affect muscle force and lever arms change throughout a muscle's range of motion.[187] For reported data to be a useful resource for comparison with tests of individuals, therefore, the subject position used during the collection of such data requires adequate documentation.

Subject Stabilization During Testing

Optimum subject stabilization for testing has received little attention. Available information appears to be limited to tests of muscles around the knee joint. Hanten and Ramberg[140] compared thigh, pelvic, and trunk straps with no straps during concentric and eccentric knee extension tests. They found no difference between measurements obtained under maximum and minimum stabilization. In a similar study by Hart et al[174] of concentric knee extension, force measurements were found to increase when pelvic and trunk straps were added to thigh straps. Hanten and Ramberg[140] tested female subjects, whereas Hart et al[174] tested male subjects. Of perhaps greater importance, Hanten and Ramberg had subjects in both groups (stabilized maximally or minimally) grip the sides of the chair during testing, whereas Hart et al did not instruct their subjects to do so. Gripping the sides of the chair may have afforded similar stability to that offered by strapping. Patteson et al[188] evaluated the effect of strapping the contralateral thigh during knee flexion and extension tests. Subjects were asked to grip the sides of the chair during testing, and again no differences were found between measurements obtained under those conditions. Because so little can be confidently deduced regarding the influence of subject stabilization on force measurements, it should be assumed that different forms of subject stabilization have the potential to influence the magnitude of obtained values

Method Used for Axes Alignment

Alignment of joint and lever arm axes is required for interpretation of forces applied to the lever arm. When the axes are aligned, the limb and the machine act upon each other with the same moment arm.[104] If alignment of axes varies, lever arm lengths may differ.[189] Although most authors claim to align equipment and joint axes, methods used vary. For knee tests, the axis of the dynamometer has frequently been reported to be visually aligned with the axis of the knee joint.[87,132,140,182,188] Such a definition provides little guidance to aid replication of the procedure. We believe that it is more appropriate to define the anatomical landmark used to represent the changing axis of knee joint movement. Some authors have aligned the axis of the dynamometer with the lateral epicondyle,[52,152,159,184,190] a horizontal line through the femoral condyles,[103,191] or a point near the lateral epicondyle where a minimum of slippage of the resistance pad against the tibia occurs as the knee flexes and extends through a 90-degree arc of motion.[192] Others have failed to specify the subject axis.[5,102,193,194]

If the axis of the dynamometer is aligned with the lateral epicondyle when the knee is resting in 90 degrees of flexion, the axes can be several centimeters out of alignment when the knee is moved to full extension. In general, because research reports fail to include reports of the knee joint angle at which alignment was established, replication of the alignment procedures is not possible. An even more complex problem arises when attempts are made to align the axis of the dynamometer with multiaxial joints such as the shoulder joint. How the joint axis is approximated and the position of the joint when axes are aligned should therefore be reported.

Determining Lever Arm Length

When knee extension is tested, the axis of the lever arm is aligned with an estimated axis of the knee. The resistance pad on the lever arm is then attached to the tibia. Methods used to position the resistance pad and to determine lever arm length vary.[103,157] In addition, the criterion used for determination of lever arm length is commonly not specified.[4,39,146,151,195]

The force applied to the transducer will vary with distance from the axis of the knee joint. The farther from the joint axis the transducer is placed, the less should be the force registered by the transducer during a maximal contraction. The equation force x distance should, however, produce a constant value for the torque that moves the tibia independent of the position of the resistance pad along the tibial lever arm. Thus, altering the location of the resistance pad should not, theoretically, affect torque.

Several authors have reported that torque is not independent of lever arm length. Otis and Gould[192] observed that during isometric knee extension tests using the Cybex[R] II dynamometer,([double dagger]) ([double dagger]) the farther from the knee joint the resistance pad was placed, the greater were the torques recorded. Although Otis and Gould failed to correct torques for the effect of gravity, it is unlikely that this could account for the torque decrements associated with diminished lever arm length. At 90 degrees of flexion, the effect of gravity on torques would be negligible, and torque changes associated with altered lever length appeared to be similar regardless of whether knee extension was performed at 90 or 30 degrees of knee flexion. In addition, although the reliability of the measurements was not reported, systematic effects on torque that were attributable to lever arm length were demonstrated. Otis and Gould argued that the observed differences occurred because the torque generated around the knee is determined by the distance of the patellar tendon from the changeable axis of the knee joint. As this distance is influenced by the relative position of the tibia and the femur, differences in resistance-pad location that move the tibia relative to the femur would alter the effective knee torque despite forces applied through the patellar tendon being the same. Cadaveric investigations cited by Otis and Gould provide evidence that the magnitude of possible differences in location of the knee axis are sufficient to account for the observed torque differences.

The findings reported by Kramer et al[196] support the observation that torque is influenced by lever arm length. Using the Kin-Com[R] dynamometer,([sections]) ([sections]) they measured the torque during concentric-eccentric knee extension at 60 [degrees] /s using dynamometer arm lengths corresponding to 33%, 67%, and 95% of the distance from the estimated location of the knee axis to a resistance-pad placement that just contacted the dorsum of the foot. The torques produced at the 33% length were approximately 39% lower than those produced at the 67% length (or approximately 50%-55% of the 95% length torques), and the torques produced at the 67% length were approximately 10% lower than those produced at the 95% length (approximately 86%-90% of the 95% length torques). Kramer and colleagues hypothesized that these differences were due to the effect of lever arm length on alignment of the axes of the knee and the dynamometer, alteration in the angle of the tibia relative to the horizontal at different lever arm lengths, and subject inhibition due to discomfort as the resistance pad is placed higher on the tibia. No evidence was provided, however, that these hypothesized mechanisms could account for the large differences in torque found for the three lever arm lengths used. In addition, no attempt was made to justify disregarding the explanation provided earlier by Otis and Gould.[192] Taylor and Casey,[197] using the Cybex[R] II dynamometer, compared torques produced at 25% (5-10 cm [2-4 in]) or greater shortening of the lever arm with torques produced at maximum usable leg length. The results again support the argument that lever arm length influences torque determinations. Taylor and Casey recommended placing the resistance pad at the most distal usable leg length and disregarding the manufacturer's statement that torque production is independent of lever arm length.

For the conditions studied, shortening the lever arm appears to result in decreased torque recordings. Whether the torque measured for muscles that cross the knee joint may be affected in a meaningful way by small alterations (eg, 1-3 cm) of lever arm length such as may occur clinically due to inconsistencies in axes alignment or resistance-pad placement, however, remains to be determined. Twenty-five percent reductions in usable leg length affect the magnitude of torque measurements, but smaller reductions have not been reported. In addition, the proposed explanations for this effect have not been adequately studied to determine the relative influence of each explanation on observed torque differences. Furthermore, how methods used to determine lever arm length affect torque measurements when joints other than the knee are tested have not been investigated. Standardization and reporting of the technique used for determining lever arm length, therefore, appear to be necessary if systematic differences in torque measurements associated with this aspect of test setup are to be avoided.

Determining Preload

The preload or activation force is a preselected minimum force that must be applied to the load cell to initiate movement of the lever arm. The purposes of the preload are to prevent accidental initiation of lever arm movement and to allow buildup of force generated by the tested muscles so that maximum torque is achieved earlier in the test range of movement than under conditions of no preload. As test speed increases and the time to move through the test range of movement and achieve peak torque decreases, it is thought that the need for preload increases. Although measurements of peak torque appear to be unaffected by the presence or absence of preload, the absolute value of the preload has been reported to affect average torque measurements.[152,190] Reports[170,198,199] indicate that torque measurements in the initial portion of test movement increase as preload increases. This increase would seem to be predictable as those forces that are applied prior to achievement of the preload force are no longer recorded. If an average of torque recordings over the test range of movement is needed, the obtained value could therefore be expected to increase as preload increases. Consequently, Kramer et al[152] suggested that comparisons of torques should be based on comparable preload forces.

Although the optimal method for determining preload has not been addressed, three methods have been reported:

1. The same preload is used for all subjects tested. The magnitude of preload varies with the joint and movement tested. Examples for tests of knee extension include 25 N[200] 50 N,[33,52,196] and 150 N.[201]

2. Preload is determined for each subject using a percentage of the maximum torque the subject is capable of generating under the test conditions. The obtained torque curve then arguably represents a similar proportion of that subject's effort.

3. Preload is a multiple of limb weight. Kues et al[184] used a preload that was 150% of the passive weight of the lower limb. This preload was apparently adequate to prevent accidental initiation of lever arm movement.

A logical argument for one preload determination method over another has not been made in the literature and warrants attention. The magnitude of preload has been shown to affect the magnitude of measurements that are averaged over the test range of movement. When protocol replication is desirable, the preload used should also be replicated and should be reported when protocol descriptions are published.

Choice of Damp/Ramp Settings

During the initial arc of motion tested, the lever arms of some electromechanical dynamometers accelerate to reach the preset speed. The point at which the preset speed is reached may show a bump in the torque curve, commonly referred to as "torque overshoot."[104] Sapega et al[202] reported that overshoot can result in erroneously high torque measurements that occur as the machine arrests the inertia developed by the accelerating limb. These errors increase when large limb segments are tested, large amounts of torque are developed, or small ranges of movement are tested. The problem of torque overshoot has been dealt with by using a window of data,[203] damping of data,[104] and acceleration ramping.[204]

Using a window of data carries the limitation that, at high test speeds, torque overshoot oscillations may continue late into the range of movement.[202] Damping electronically modifies signals received such that torque recordings are filtered to minimize oscillations. The Cybex[R] II dynamometer allows damp settings from 0 (no damp) to 4 (maximum damp). Damp settings can have a dramatic effect on the magnitude of torque measurements. In addition, as damp increases, the recorded torque curve is shifted to the right.[104] If angle-specific measurements were the criterion used, the magnitude of such measurements would therefore be influenced by the damp selected. The Kin-Com[R] dynamometer does not have damp settings, but torque overshoot is modified by acceleration and deceleration rampings that can be set at low, medium, or high. These ramp settings put a ceiling on the amount of acceleration of the lever arm allowed by the instrument.

Rathfon et al[204] examined the effects of different acceleration and deceleration rates on torques produced by knee extensors tested at 90 [degrees] /s. They concluded that the choice of ramping did not appear to have a meaningful effect on torque averaged over the whole curve nor did it affect peak torque. Their findings, however, must be confined to the protocol studied. Ramps selected for tests of stronger subjects, heavier body segments, smaller arcs of movement, or higher speeds may affect data differently. Rathfon et al[204] emphasized that the size of the window of data that represents constant lever arm speed varies with ramp setting. In their experiment, higher ramp settings resulted in a decrease in the test range of movement that occurred at constant lever arm speed. Higher rampings apparently resulted in greater fluctuations in lever arm speed before a constant speed was reached. They concluded that the practice of deleting a predetermined arc of motion from the beginning and end of the torque curve to eliminate acceleration and deceleration phases is not justified unless verified for the protocol in use. Thus, damping can affect the magnitude of the torque curve and of angle-specific measurements, and ramping can alter the size of the isokinetic window. Damp and ramp settings, therefore, should be reported when results are published.

Gravity-Correction Procedures

The importance of considering the influence of gravity on dynamometric measurements has been argued by several authors.[104,158,178,205] Gravity-correction procedures increase measurements for movements against gravity and decrease measurements for movements with gravity. Gravity correction is needed if data will subsequently be used to form ratios.[206] Corrected values theoretically provide a better estimation of torque generated by muscles than do uncorrected values.

Gravity correction of measurements is most commonly accomplished in one of two ways. Limb weight at a single point in the test range of movement can be used to estimate limb weight at the horizontal. The weight of the limb at various points through range of movement can then be determined. These estimates of correction values are made based on the assumption that the passive weight of the limb acting on the transducer will be related to limb weight at the horizontal by the cosine of the angle of the limb to the horizontal. Alternatively, direct measurement of limb weight can be obtained as the limb is moved passively through the test range of movement. Although van der Leeuw et al[158] have demonstrated that passive forces acting during knee flexion and extension are linearly, not cosinally, related to joint angle, the effect on data due to selection of correction method has not been established. Nevertheless, it is common for authors to report having used gravity-correction procedures on their data, without indicating the method used.([parallel]) ([parallel])

If gravity were the sole source of passive restraint to knee extension, the passive torque curve for knee flexion and extension would be cosinal in shape. The fact that it is linear[158] suggests that other structures that attach to the tibia, such as the hamstring muscles and structures posterior to the knee joint, may be placed under increasing tension as the knee extends, resulting in forces being registered by the transducer that are not due to the effect of gravity alone. The recent work by Finucane et al[206] confirms that the position in the range of movement selected for weighing the legs will influence the correction values applied to measurements of knee extension. It is possible that this is a consequence of passive forces not attributable to gravity that act over the test range of movement.

Our investigations indicate that the method used to determine the angle of the limb relative to the horizontal at the position selected for weighing affects the magnitude and shape of the cosine curve that is subsequently constructed. Although the Kin-Com[R] user's manual advises "for correct gravity-correction calculations, it is assumed that the reference value (relationship to horizontal at the position selected for weighing) is correct," no information regarding how this angle should be determined is provided. Many researchers,[137,157,158,196,213,214] although not specifying the method used to determine the angle of the limb relative to the horizontal, imply that they use the joint angle as a measure of this angle. Because the joint angle represents the relationship between the tibia and the femur rather than the lower limb and the horizontal, this does not appear to be a logical practice. Other authors[123,177,190,204,206] appear to base estimates on the angle of the lever arm relative to the horizontal.

Given the potential for variability in the value used for the cosine curve construction, it appears that the best method for correcting measurements of passive forces that oppose or assist active contractions would be to measure the passive torque through the range of movement for each subject and then correct the active torque measurement at each angle using the passive torque measurement obtained at the same angle. Such corrections would include all passive forces registered by the transducer at different parts of the test range of movement and are not influenced by errors associated with the estimation of limb position relative to the horizontal at the point selected for weighing. Because these corrections involve manipulation of large volumes of data, however, they appear unlikely to be adopted by clinicians.

Until more information is available, whether data are corrected for the effect of gravity and the method used for these corrections should be reported. When cosine corrections are used, details should include the position selected for weighing and how limb position relative to the horizontal during weighing is determined.

Test Conditions

Test Speed

There is abundant evidence that the speed used for isokinetic testing influences the magnitude of torque measured.(##) Concentrically measured torque decreases as test speed increases.(***) Thus, obtained measurements should be considered speed-specific, and comparison of measurements should be restricted to those obtained at comparable test speeds.

Torques measured eccentrically do not appear to demonstrate as consistent a relationship to test speed. For eccentric knee extension, forces measured at 30 [degrees], 120 [degrees], and 270 [degrees] /s have been reported to not differ significantly.[144] Similarly, the data provided by Hanten and Ramberg[140] indicate that eccentric knee extension force varies little and randomly with test speed. Eccentric elbow extension force also appears to be unaffected by test speed.[171] In contrast, eccentric elbow flexion force has been reported to increase as test speed increases,[70,171,187]

Which speed is selected for testing may be of clinical importance. Differences between measurements obtained from professional and amateur athletes have been found to be significant at low but not at higher test speeds.[11,62] Appen and Duncan[134] found no difference between hamstring/quadriceps femoris muscle ratios for distance runners and sprinters except at a test speed of 300 [degrees] /s, when sprinters had a higher ratio than did distance runners. Kannus et al[217] found that subjects with 8-year-old partial anterior cruciate ligament tears showed decreased measurements for flexion of the injured limb compared with the uninjured limb, but only when tested at higher speeds. In a study comparing the strength of the injured limb with that of the contralateral uninjured limb, Kannus and Jarvinen[217] found that the relative strength deficit of the hamstring muscles of the injured limb increased with increased test speeds. Similarly, Prietto et al[43] found that impairment was speed-specific for knee extension torque in subjects with partial menisectomy, with the greatest apparent deficit occurring at the highest test speeds.

The speed used for testing has an important influence on the magnitude of force measurements. Comparisons between independently obtained sets of force measurements can only be justified for concentric tests at comparable speeds. Furthermore, conclusions based on force measurements obtained at a particular speed may not represent the conclusions drawn from force measurements obtained at a different speed.

Rest Intervals During Testing

The rest interval separating test repetitions may influence force measurement. When a rest interval of 30 seconds interrupted reciprocal knee flexion and extension test repetitions, the torques produced were on average 5% greater than when test repetitions were performed without rests.[219] When no rest separates test repetitions, measurements may show a declining linear or curvilinear trend.[66,219,220]

The effect of no rest interval on the magnitude of forces may differ depending on whether muscles are tested eccentrically or concentrically. Gray and Chandler[181] tested subjects with 40 consecutive concentric or eccentric quadriceps femoris muscle contractions. Peak torque production for the eccentric tests was found to decline by an average of 0.3%, whereas peak torque production for the concentric tests declined 47.7%. These findings are supported by Tesch et al.[71]

The way in which subjects fatigue when tested with a no-rest protocol may be partially attributed to individual differences in muscle composition. Individuals with a high fast-twitch fiber composition appear to fatigue more and recover less than individuals with a high proportion of slow-twitch fibers.[76] Individual strength differences may also affect fatigue curves. Patton et al[220] found that male subjects with high strength fatigued most rapidly and female subjects with low strength fatigued least rapidly. Therefore, the rest interval requirements for stable data may vary for individual subjects.

Thus, it appears that no rest between repeated tests of the same movement results in diminishing test measurements. The magnitude of peak and averaged torque measurements, therefore, may be influenced by the rest interval provided between repetitions.

Type of Feedback Given to Subjects

Feedback to subjects about their force production has been reported to affect measurements. Hald and Bottjen[105] found that visual feedback increased knee flexion and extension forces measured at 60 [degrees] and 180 [degrees] /s. Although measurement reliability was not established, systematic effects attributable to the provision of feedback were not concealed by the measurement error, however large. Feedback was given on either the first or second test occasion, which would appear to invalidate systematic effects attributable to occasion as a competing explanation for the increases. Although subjects produced greater torques with feedback, the differences appeared to be small. Baltzopoulos et al[221] noted increases in gravity-corrected knee flexion and extension forces when visual feedback was provided. The forces attained with feedback were greater when the test was conducted at 60 [degrees] /s but not at 180 [degrees] /s. The small sample size (n=10) may account for the apparently greater forces failing to attain statistical significance when feedback was used at 180 [degrees] /s. Figoni and Morris[222] found that visual feedback given to 20 male subjects resulted in an approximately 12% increase in peak torque for knee flexion and extension tested at 15 [degrees] /s, but no differences were found at 300 [degrees] /s. Although the reliability of the measurements was not reported, the randomized application of feedback or no-feedback conditions over the two test occasions probably excluded systematic effects due to occasion or sequence from explaining the observed differences. The actual percentage of increase reported may not be particularly useful because the data were not corrected for the effects of gravity. Percentage improvements associated with feedback for measurements that have not been gravity corrected will not necessarily reflect those that might occur when gravity correction is applied.

Peacock et al[223] found that visual and auditory feedback enhanced knee extension force measurements. The authors did not report the test speed that was used, whether data were gravity corrected, or evidence of reliability of measurements. No systematic trends attributable to test occasion were noted. These limitations make the-observed increases of approximately 10% difficult to interpret or apply. The random application of feedback or no-feedback conditions, however, lend credibility to the authors' conclusion that feedback was responsible for the observed increases.

Despite the various limitations of these reports, the evidence suggests that, at least for slow test speeds, feedback is likely to enhance force production. It is nevertheless the more common practice for authors to fail to reveal whether subjects were provided with feedback. As another potential source of systematic measurement variability, it would appear appropriate that the presence and nature of feedback should be documented when test procedures are described.

Test Repetitions

How many test repetitions are performed and which repetitions are selected for analysis may also influence the magnitude of force measurements. Murray et al[7] found that average torque was higher on the second of two consecutive isokinetic trials. Johnson and Siegel[116] found a linearly increasing trend in the first three test repetitions. Stratford[149] found a similar linear trend for knee extension peak torque measurements for both the injured and uninjured limbs of patients with anterior cruciate ligament reconstruction. Burdett and Swearingen[132] reported that peak values for knee flexion and extension tests were not reached until the second or third repetition. Kues et al,[184] based on visual analysis of data for knee tests, concluded that subjects should be tested using four maximal repetitions and the greatest measurement obtained should be used to represent the subjects' maximal effort. The importance of defining the number of test repetitions used is presumably accepted as it is one of the more consistently reported aspects of experimental procedure.

Whether data from the first or subsequent test occasions are considered also appears to affect the magnitude of obtained force measurements. Several authors[7,193,224] report that measurements obtained on a second occasion were higher than those obtained on the first occasion. The fact that measurements have been demonstrated to systematically vary with test occasion underlines the importance of establishing temporal stability of data.

Data Analysis

Several performance measures have been presented in the literature. The most common of these measures are the maximum (peak) torque produced from a series of repetitions, peak torque averaged across repetitions, the average torque produced during a single repetition or averaged across repetitions, average or maximum work and power, and angle-specific torque (torque selected at a specified joint angle). Experimental conclusions based on one particular type of measurement may not necessarily apply if other kinds of measurements are used. To ensure reproducibility of test procedures, therefore, it is important to clearly describe the measures used and how they were manipulated in the data analysis.

Conclusion and Recommendations

The review indicates that dynamometric measurements are affected by many factors that are an unavoidable part of test procedure. Movement-related variables, pretest procedures, test conditions, data-analysis methods, and subject factors (Table) were all found to affect force measurements. These systematic effects indicate cause for concern over the exact procedure applied to collect data. In this respect, our analysis confirms and extends issues raised by previous reviews.[96,97]

Checklist for Reporting Test Procedures

Subject-related factors

Age

Gender

Weight

Athletic background

Disability

Dominance

Test-related factors

Pretest

Warm-up procedures

Subject starting position

Stabilization

Alignment of anatomical axis and dynamometer axis

Lever arm length

Preload

Choice of ramp, damp settings

Gravity-correction method and whether it was used

During Test

Speed

Rest intervals

Feedback

Joint angle/range of motion tested

Muscle action (concentric, eccentric)

Mode (isokinetic, isometric, isotonic)

Posttest 

Type of data

Data analysis

These issues are pertinent to a wide range of research and clinical applications: when normative data are reported, when ratios are generated, and when measurements are correlated to other criteria or compared across studies. Clinicians who apply normative data or ratios to tests of an individual must be able to replicate the protocols used in normative studies. Similarly, comparison of measurements for repeated tests of an individual requires replication of test conditions if confounded interpretation is to be avoided. In addition, the reliability of clinical assessment cannot be assumed to be comparable to that published, unless protocol variables are comparable.

Yet, despite more than a decade of publications indicating systematic biases in test findings due to particular protocol variations, the majority of research publications in this field do not provide sufficient protocol or subject descriptions to allow replication of test procedures. Pitetti,[212] for example, claimed that the protocol described for testing elbow and knee flexion and extension could be used for job assessment and rehabilitation purposes. The protocol description, however, omitted several aspects shown to influence data: how gravity correction was performed, the damping used, the preload used, how axes were aligned, how lever arm length was determined, whether feedback was given to subjects, the range of joint movement tested, what kind of rest separated repetitions, and the isokinetic window used for data analysis.

Factors that influence force measurements were identified when change in the test condition under examination produced differences in test results. However, when no differences in test results were reported despite alteration to test conditions, the conclusion about the influence of the factor in question is equivocal. The lack of meaningful differences indicates either similarity of measurements under the compared test conditions or statistical insensitivity. Such insensitivity increases as sample size diminishes, and small sample sizes were a common feature of the publications reviewed. The insensitivity also increases as random variability in obtained measurements increases. This random variability could be influenced by experimental method. In this respect, it is important to note that the majority of publications examined omitted either essential information about test procedures or information about the magnitude of measurement error. A major implication of the arguments presented, therefore, is that protocol and subject factors need more extensive documentation than is typical at present. In general, it seems reasonable to suggest that until additional information is made available, at least those factors listed in the Table should be reported.

Another major implication is that some standardization of test methods used may prove advantageous, but further research is needed to guide that process. As this review showed, in current practice test procedures vary considerably, making it difficult to determine why experimental conclusions conflict. Many of the research questions that remain unanswered, such as the effect of gender on weight-adjusted measurements or the relationship between height and measurements for children, might have been resolved by now if comparison of results had been facilitated by similar experimental procedures. Of great additional importance, data on subjects with impairments, which are often difficult to collect in large volumes, could be collated to form a more representative database if experimental conditions under which data were collected were standardized.

Standardization of protocols is dependent on knowledge about those factors that influence measurements. This review indicates the need for further research into the factors that influence measurements. Although the literature has already identified many test factors that appear to systematically influence measurements, additional information is needed regarding optimal test procedures. Outstanding questions identified by this review include the way in which the method used for aligning axes affects measurements, how lever arm length and preload should be determined, the influence on measurements when different methods for gravity correction are used, optimum rest intervals and warm-up requirements, and whether feedback should be used. Additional research on how measurements are affected by subject factors such as age, weight, and determination of lower-limb dominance is also indicated. In particular, it may be necessary to obtain norms for specific subgroups of subjects if normative data are to be used to interpret the results of tests of individuals in the clinic.

([double dagger]) References 4, 14, 21, 25, 33, 52, 58, 103, 155. ([sections]) References 14, 21, 25, 33, 52, 58, 103, 142, 155. ([parallel]) References 1, 21, 49, 56, 71, 128, 144, 163, 164. (#) References 2, 15, 17, 24, 25, 31, 47, 48, 59, 65, 73, 74, 81, 89, 91, 102, 167-173. (**) References 21, 52, 116, 128, 138, 140, 175, 176. ([dagger])([dagger]) References 16, 89, 102, 118, 128, 130, 145, 185, 186. ([double dagger])([double dagger]) Cybex, Div of Lumex Inc, 2100 Smithtown Ave, Ronkonkoma, NY 11779. ([subsections]) Chattecx Corp. 101 Memorial Dr, PO Box 4287, Chattanooga, TN 37405. ([parallel])[parallel]) Refeences 8, 15, 21, 56, 78, 122, 127, 138, 145, 164, 166, 170, 182, 194, 207-212. (##) References 11, 16, 17, 26, 51, 57-62, 124, 132, 137, 138, 143, 152, 153, 164, 166, 180, 215, 216. (***) References 2, 8, 14, 49, 58, 62, 65, 70, 87, 128, 139-145.

References

[1] Alexander MJ. Peak torque values for antagonist muscle groups and concentric and eccentric contraction types for elite sprinters. Arch Phys Med Rehabil. 1990;71:334-349.

[2] Cahalan TD, Johnson ME, Liu S, Chao KY. Quantitative measurements of hip strength in different age groups. Clin Orthop. 1989;246:136-145.

[3] Falkel J. Plantar-flexor strength testing using the Cybex isokinetic dynamometer. Phys Ther. 1978;58:847-850.

[4] Goslin BR, Charteris J. Isokinetic dynamometry: normative data for clinical use in lower extremity (knee) cases. Scand J Rehabil Med. 1979;11:105-109.

[5] Highenboten CL, Jackson AW, Meske NB. Concentric and eccentric torque comparisons for knee extension and flexion in young adult males and females using the kinetic communicator. Am J Sports Med. 1988;16:234-237.

[6] Mayer T, Barnes D, Nichols G, et al. Progressive inertial lifting evaluation, 11: a comparison with isokinetic lifting in a disabled chronic low-back pain industrial population. Spine. 1988;13:998-1002.

[7] Murray PM, Gardner GM, Mollinger LA, Sepic SB. Strength of isometric and isokinetic contractions. Phys Ther. 1980;64:412-419.

[8] Schlinkman B. Norms for high-school football players derived from Cybex data reduction computer. J Orthop Sports Phys Ther. 1984;5:243245.

[9] Seeds R, Levene J, Goldberg H. Normative data for Isostation B100. J Orthop Sports Phys Ther. 1987;9:141-155.

[10] Seeds R, Levene J, Goldberg H. Abnormal patient data for the Isostation B100. J Orthop Sports Phys Ther. 1988;10:121-133.

[11] Smith DJ, Quinney HA, Wenger HA, et al. Isokinetic torque output of professional and elite amateur ice hockey players. J Orthop Sports Phys Ther. 1981;3:42-47.

[12] Smith SS, Mayer TG, Gatchel RJ, Becker TJ. Quantification of lumbar function, part 1: isometric and multispeed isokinetic trunk strength measures in sagittal and axial planes in normal subjects. Spine. 1985;10:757-764.

[13] Timm KE. Isokinetic lifting simulation: a normative data study. J Orthop Sports Phys Ther. 1988;10: 156-166.

[14] Weltman A, Tippett S, Janney C, et al. Measurement of isokinetic strength in prepubertal males. J Orthop Sports Phys Ther. 1988;9:345351.

[15] Harter RA, Osternig LR, Standifer LW. Isokinetic evaluation of quadriceps and hamstrings symmetry following anterior cruciate ligament reconstruction. Arch Phys Med Rehabil. 1990;71:465-468.

[16] Hinton RY. Isokinetic evaluation of shoulder rotational strength in high-school baseball pitchers. Am J Sports Med. 1988;16:274-279.

[17] Kannus P, Jarvinen M. Knee flexor/extensor strength ratio in follow-up of acute knee distortion injuries. Arch Phys Med Rehabil. 1990;71:38-41.

[18] Mira AJ, Carlisle KM, Greer RB. A critical analysis of quadriceps function after femoral shaft fracture in adults. J Bone Joint Surg [Am]. 1980;62:61-67. [19] Chmelar RD, Shultz BB, Ruhling RO, et al. Isokinetic characteristics of the knee in female professional and university ballet and modern dancers. J Orthop Sports Phys Ther. 1988;9:410-418.

[20] Mayer TG, Kondraske G, Mooney V, et al. Lumbar myoelectric spectral analysis for endurance assessment: a comparison of normals with deconditioned patients. Spine. 1989;14:986-991.

[21] Dvir Z, Shklar A, Halperin N, et al. Concentric and eccentric torque variations of the quadriceps femoris in patellofemoral pain syndrome. Clin Biomech. 1990;5:68-72.

[22] Armstrong LE, Winant DM, Swasey PR, et al. Using isokinetic dynamometry to test ambulatory patients with multiple sclerosis. Phys Ther. 1983;63:1274-1279.

[23] Bennett JG, Stauber WT. Evaluation and treatment of anterior knee pain using eccentric exercise. Med Sci Sports Exerc. 1986;18:526-530.

[24] Carpenter DM, Graves JE, Pollock ML, et al. Effect of 12 and 20 weeks of resistance training on lumbar extension torque production. Phys Ther. 1991;71:580-588.

[25] Grace TG, Sweetser ER, Nelson MA, et al. Isokinetic muscle imbalance and knee-joint injuries. J Bone Joint Surg [Am]. 1984;66:734-740.

[26] Hortobagyi T, Katch FI. Eccentric and concentric torque-velocity relationships during arm flexion and extension: influence of strength level. Eur J Appl Physiol. 1990;60:395-401.

[27] Onuoha ARA. Isokinetic testing and training with the Cybex II+ dynamometer. Physiotherapy. 1990;76:731-733.

[28] Stanton P, Purdam C. Hamstring injuries in sprinting: the role of eccentric exercise. J Orthop Sports Phys Ther. 1989;10:343-349.

[29] Trudell-Jackson E, Meske N, Highenbothen C, Jackson A. Eccentric/concentric torque deficits in the quadriceps muscle. J Orthop Sports Phys Ther. 1989;11:142-145.

[30] Williamson SC, Hartigan C, Morgan RF, et al. Computerized analysis of isokinetic torque curves for muscle strengthening. J Burn Care Rehabil. 1989;10:160-166.

[31] Agre JC, Pierce LE, Raab DM, et al. Light resistance and stretching exercise in elderly women: effect upon strength. Arch Phys Med Rehabil. 1988;69:273-276. [32] Anderson AF, Lipscomb AB. Analysis of rehabilitation techniques after anterior cruciate reconstruction. Am J Sports Med. 1989;17:154160. [33] Duncan PW, Chandler JM, Cavanaugh DK, et al. Mode and speed specificity of eccentric and concentric exercise training. J Orthop Sports Phys Ther. 1989;11:70-75.

[34] Ellenbecker TS, Davies GJ, Rowinski MJ. Concentric versus eccentric isokinetic strengthening of the rotator cuff: objective data versus functional test. Am J Sports Med. 1988;16:64-69.

[35] Gerberich G, Erickson D, Serfass R, et al. Quadriceps strength training using two forms of bilateral exercise. Arch Phys Med Rehabil. 1989;70:775 779.

[36] Jensen K, Di Fabio RP. Evaluation of eccentric exercise in treatment of patellar tendinitis. Phys Ther. 1989;69:211-216.

[37] Knuttson E, Skoglund CR, Natchev E. Changes in voluntary muscle strength, somatosensory transmission, and skin temperature concomitant with pain relief during autotraction in patients with lumbar and sacral root lesions. Pain. 1988;33:173-179.

[38] Komi PV, Buskirk ER. Effect of eccentric and concentric muscle conditioning on tension and electrical activity of human muscle. Ergonomics. 1972;15:417-434.

[39] Lesmes GR, Costill DL, Coyle EF, Fink WJ. Muscle strength and power changes during maximal isokinetic training. Med Sci Sports. 1978;10:266-269.

[40] McMullen W, Roncarati A, Koval P. Static and isokinetic treatments of chondromalacia patella: a comparative investigation. J Orthop Sports Phys Ther. 1990;12:256-266.

[41] Nelson AG, Arnall DA, Loy SF, et al. Consequences of combining strength and endurance training regimens. Phys Ther. 1990;70:287294.

[42] Pavone E, Moffat M. Isometric torque of the quadriceps femoris after concentric, eccentric, and isometric training. Arch Phys Med Rehabil. 1985;66:168-170.

[43] Prietto CA, Caiozzo VJ, Prietto PP, McMaster WC. Closed versus open partial meniscectomy: postoperative changes in the force-velocity relationship of muscle. Am J Sports Med. 1983;11:189-194.

[44] Rabin SI, Post M. A comparative study of clinical muscle testing and Cybex evaluation after shoulder operations. Clin Orthop. 1990;258:147-156.

[45] Aronen JG, Regan K Decreasing the incidence of recurrence of first-time anterior shoulder dislocations with rehabilitation. Am J Sports Med. 1984;12:283-291. [46] Bishop KN, Durrant E, Allesen PE, Merril G. The effect of eccentric strength training at various speeds on concentric strength of the quadriceps and hamstring muscles. J Orthop Sports Phys Ther. 1991;13:226-230.

[47] Cote C, Simoneau J, Lagasse P, et al Isokinetic strength training protocols: Do they induce skeletal muscle fiber hypertrophy? Arch Phys Med Rehabil. 1988;69:281-285.

[48] Esselman PC, De Lateur BJ, Alquist AD, et al. Torque development in isokinetic training. Arch Phys Med Rehabil. 1991;72:723-728.

[49] Kramer JF, MacDermid J. Isokinetic measures during concentric-eccentric cycles of the knee extensors. Australian Journal of Physiotherapy. 1989;35:9-14.

[50] Petersen SR, Bagnall KM, Wenger HA, et al. The influence of velocity specific resistance training on the in-vivo torque-velocity relationship and the cross sectional area of the quadriceps femoris. J Orthop Sports Phys Ther. 1989;10:456-462.

[51] Griffin JW. Differences in elbow flexion torque measured concentrically, eccentrically, and isometrically. Phys Ther. 1987;67:1205-1208.

[52] Hageman PA, Gillaspie DM, Hill DH. Effects of speed and limb dominance on eccentric and concentric isokinetic testing of the knee. J Orthop Sports Phys Ther. 1988;10:59-65.

[53] Knapik JJ, Roberta MS, Mawdsley RH, Ramos M. Angular specificity and test mode specificity of isometric and isokinetic strength training. J Orthop Sports Phys Ther. 1983;5:58-65.

[54] Knapik JJ, Wright J, Mawdsley RH, Braun J. Isometric, isotonic, and isokinetic torque variations in four muscle groups through a range of motion. Phys Ther. 1983;63:938-947.

[55] Petersen S, Wessel J, Bagnall K, et al. Influence of concentric resistance training on concentric and eccentric strength. Arch Phys Med Rehabil. 1990;71:101-105.

[56] Vyse WM, Kramer JF. Interaction of concentric and eccentric muscle actions during continuous activation cycles of the elbow flexors. Physiotherapy Canada. 1990;42:123-127.

[57] Caiozzo VJ, Perrine JJ, Edgerton VR. Training-induced alterations of the in vivo force-velocity relationship of human muscle. J Appl Physiol. 1981;51:750-754.

[58] Costain R, Williams AK. Isokinetic quadriceps and hamstring torque levels of female adolescent soccer players. J Orthop Sports Phys Ther. 1984;5:196-200.

[59] Figoni SF, Christ CB, Massey BH. Effects of speed, hip and knee angle and gravity on hamstring to quadriceps torque ratios. J Orthop Sports Phys Ther. 1988;9:287-291.

[60] Gilliam TB, Sady SP, Freedson PS, Villanacci J. Isokinetic torque levels for high-school football players. Arch Phys Med Rehabil. 1979;60: 110-114.

[61] Kanehisa H, Miyashita M. Specificity of velocity in strength training. Eur J Appl Physiol 1983;52:104-106.

[62] Alway SE, Stray GJ, Grumbt WH, Gonvea WJ. Muscle cross-sectional area and torque in resistance-trained subjects. Eur J Appl Physiol 1990;60:86-90.

[63] LoPresti C, Kirkendall DT, Street GM, Dudlet AW. Quadriceps insufficiency following repair of the anterior cruciate ligament. J Orthop Sports Phys Ther. 1988;9:245-249.

[64] Mayer TG, Vanharanta H, Gatchel RJ, et al. Comparison of CT scan muscle measurements and isokinetic trunk strength in postoperative patients. Spine. 1989;14:33-36.

[65] Barnes W. The relationship of motor unit activation to isokinetic muscular contraction at different contractile velocities. Phys Ther. 1980;60:1152-1158.

[66] Elert J, Gerdle B. The relationship between contraction and relaxation during fatiguing isokinetic shoulder flexions: an electromyographic study. Eur J Appl Physiol. 1989;59:303-309.

[67] Hinson M, Rosentswieg J. Comparative and electromyographic values of isometric, isotonic, and isokinetic contraction. Res Q. 1975; 44:71-78.

[68] Klein AB, Snyder-Mackler L, Roy SH, DeLuca CJ. Comparison of spinal mobility and isometric trunk extensor forces with electromyographic spectral analysis in identifying low back pain. Phys Ther. 1991;71:445-454.

[69] Komi PV, Viitasalo JT. Changes in motor unit activity and metabolism in human skeletal muscle during and after repeated eccentric and concentric contractions. Acta Physiol Scand 1977;100:246-254.

[70] Rodgers K, Berger R. Motor unit involvement and tension during maximum, voluntary concentric, eccentric, and isometric contractions of the elbow flexors. Medicine and Science Research in Sports. 1974;6:253-259.

[71] Tesch PA, Thorsson A, Colliander EB. Effects of eccentric and concentric resistance training on skeletal muscle substrates, enzyme activities, and capillary supply. Acta Physiol Scand. 1990;140:575-580.

[72] Tesch PA, Dudley GA, Duvoisini MR, et al. Force and EMG signal patterns during repeated bouts of concentric and eccentric muscle actions. Acta Physiol Scand. 1990;138:263-271.

[73] Binder-Macleod SA, Guerin T. Preservation of force output through progressive reduction of stimulation frequency in human quadriceps femoris muscle. Phys Ther. 1990;70:619-625.

[74] Ferguson JP, Blackley MW, Knight RD, et al. Effects of varying electrode site placements on the torque output of electrically stimulated involuntary quadriceps femoris muscle contraction. J Orthop Sports Phys Ther. 1989;11:24-29.

[75] Snyder-Mackler L, Garret M, Roberts M. A comparison of torque-generating capabilities of three different electrical stimulating currents. J Orthop Sports Phys Ther. 1989;10:297-301.

[76] Colliander EB, Dudley GA, Tesch PA. Skeletal muscle fiber type composition and performance during repeated bouts of maximal, concentric contractions. Eur J Appl Physiol. 1988;58:81-86.

[77] Dibrezzo R, Fort IL, Brown B. Dynamic strength and work variations during three stages of the menstrual cycle. J Orthop Sports Phys Ther. 1988;10:113-116.

[78] Einarsson G. Muscle conditioning in late poliomyelitis. Arch Phys Med Rehabil 1991;72:11-14.

[79] Einarsson G. Muscle adaptation and disability in late poliomyelitis. Scand J Rehabil Med Suppl. 1991;25:1-76.

[80] Halle JS, Smidt GL, O'Dwyer KD, Lin SY. Relationship between trunk muscle torque and bone mineral content of the lumbar spine and hip in healthy postmenopausal women. Phys Ther. 1990;70:690-699.

[81] Jansson E, Dudley GA, Norman B, Tesch PA. Relationship of recovery from intensive exercise to the oxidative potential of skeletal muscle. Acta Physiol Scand. 1990;139:147-152.

[82] MacDougall JD, Ward GR, Sale DG, Sutton JR. Biomechanical adaptation of human skeletal muscle to heavy resistance training and immobilization. J Appl Physiol: Respirat Environ Exercise Physiol. 1977;43: 700-703.

[83] Peel C, Alland MJ. Cardiovascular responses to isokinetic trunk exercise. Phys Ther. 1990;70:503-510.

[84] Tesch PA, Wright JE. Recovery from short-term intense exercise: its relation to capillary supply and blood lactate concentration. Eur J Appl Physiol 1983;52:98-103.

[85] Tesch PA, Wright JE, Vogel JA, et al. The influence of muscle metabolic characteristics on physical performance. Eur J Appl Physiol. 1985;54:237-243.

[86] Thorstensson A, Karlson J. Fatiguability and fiber composition of human skeletal muscle. Acta Physiol Scand. 1976;98:318-322.

[87] Thorstensson A, Grimby G, Karlsson J. Force velocity relations and fiber composition in human knee extensor muscles. J Appl Physiol. 1976;40:12-16.

[88] Zimmerman CL, Smidt GL, Brooks JS, et al. Relationship of extremity muscle torque and bone density in postmenopausal women. Phys Ther. 1990;70:302-309.

[89] Deutsch H, Kilani H, Moustafa E, et al. Effect of head-neck position on elbow flexor muscle torque production. Phys Ther. 1987;67:517-521.

[90] Lyon LK, Benz L, Johnson KK, et al. Q-angle: a factor in peak torque occurrence in isokinetic knee extension. J Orthop Sports Phys Ther. 1988;9:250-253.

[91] Bohannon RW, Andrews AW. Correlation of knee extensor muscle torque and spasticity with gait speed in patients with stroke. Arch Phys Med Rehabil. 1990;71:330-333.

[92] Delitto A, Lehman RC. Rehabilitation of the knee. Clin Sports Med. 1989;8:824-830.

[93] Vizsolyi P, Taunton J, Robertson G, et al. Breaststroker's knee: an analysis of epidemiological and biomechanical factors. Am J Sports Med. 1987;15:63-71.

[94] Guffey JS, Guffey DR. The reliability of selected measures of muscular performance in knee flexion and extension using the Hydra-fitness Omnitron total power unit: a pilot study. J Orthop Sports Phys Ther. 1989;10:418-422.

[95] Thompson MC, Shingleton LG, Kegerreis ST. Comparison of values generated during testing of the knee using the Cybex II Plus and Biodex Model B-2000 isokinetic dynamometers. J Orthop Sports Phys Ther. 1989;11:108-115.

[96] Mayhew TP, Rothstein JM. Measurement of muscle performance with instruments. In: Rothstein JM, ed. Measurement in Physical Therapy. New York, NY: Churchill Livingstone Inc; 1985:73-91.

[97] Dvir Z. Clinical applicability of isokinetics: a review. Clin Biomech. 1991;6:133-144.

[98] Frontera WR, Hughes VA, Lutz KJ, Evans WJ. A cross-sectional study of muscle strength and mass in 45- to 78-year-old men and women. J Appl Physiol 1991;71:644-650.

[99] Klitgaard H, Mantoni M, Sciaffino S, et al. Function, morphology, and protein expression of ageing skeletal muscle: a cross-sectional study of elderly men with different training backgrounds. Acta Physiol Scand. 1990;140:41-54.

[100] Laforest S, St-Pierre DM, Cyr J, Gayton D. Effects of age and regular exercise on muscle strength and endurance. Eur J Appl Physiol 1990;60:104-111.

[101] 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 Med Sci. 1992;47:M204-M210.

[102] Fischer NM, Pendergast DR, Calkins EC. Maximal isometric torque of knee extension as a function of muscle length in subjects of advancing age. Arch Phys Med Rehabil 1990;71:729-734.

[103] Rochcongar P, Morvan R, Jan J, et al. Isokinetic investigation of knee extensors and knee flexors in young French soccer players. Int J Sports Med. 1988;9:448-450.

[104] Rothstein JM, ed. Measurement in Physical Therapy. New York, NY: Churchill Livingstone Inc; 1985.

[105] Hald RD, Bottjen EJ. Effect of visual feedback on maximal and submaximal isokinetic test measurement of normal quadriceps and hamstrings. J Orthop Sports Phys Ther. 1987;9:86-93.

[106] Hortobagyi T, Katch FI, Katch VL, et al. Relationships of body size, segmental dimensions, and ponderal equivalents to muscular strength in high-strength and low-strength subjects. Int J Sports Med. 1990;11: 349-356.

[107] Larsson L, Grimby G, Karlsson J. Muscle strength and speed of movement in relation to age and muscle morphology. J Appl Physiol. 1979;46:451-456.

[108] Young A, Stokes M, Crowe M. Size and strength of the quadriceps muscles of old and young women. Eur J Clin Invest. 1984;14:282-287.

[109] Borkan GA, Hults SG, Gerzof AH, et al. Age changes in body composition revealed by computed tomography. J Gerontol. 1983;38: 673-677.

[110] Lexell J, Taylor CC, Sjostrom M. What is the cause of the ageing atrophy? J Neurol Sci. 1988;84:275-294.

[111] Vandervoort AA, McComas AJ. Contractile changes in opposing muscles of the human ankle joint with aging. J Appl Physiol. 1986;61: 361-367.

[112] Murray MP, Duthie EH, Gambert SR, et al. Age-related differences in knee muscle strength in normal woman. J Gerontol. 1985;40:275-280.

[113] Rice CL, Cunningham DA, Paterson DH, Rechnitzer PA. Strength in an elderly population. Arch Phys Med Rehabil 1989;70:391-397.

[114] MacLennan WJ, Hall MRP, Robinson JIT, Robinson M. Is weakness in old age due to muscle wasting? Age Ageing. 1980;9:188-192.

[115] Pearson MB, Bassey EJ, Bendall MJ. Effects of age on muscle strength and anthropometric indices within group of elderly men and women. Age Ageing. 1985;14:230-234.

[116] Johnson J, Siegel D. Reliability of an isokinetic movement of the knee extensors. Res Q. 1978;49:88-90.

[117] Nicholas JJ, Robinson LR, Logan A, Robertson R. Isokinetic testing in young nonathletic able-bodied subjects. Arch Phys Med Rehabil. 1989;70:210-2I3.

[118] Osternig LR, Bates BT, James SL. Patterns of tibial rotary torque in knees of healthy subjects. Med Sci Sports Exerc. 1980;12:195-199.

[119] Mayer T, Barnes D, Kishino N, et al. Progressive inertial lifting evaluation, 1: a standardized protocol and normative database. Spine. 1988;13:993-997.

[120] Ng L, Kramer J. Shoulder rotator torques in female tennis and nontennis players. J Orthop Sports Phys Ther. 1991;13:40-46.

[121] Oberg B, Moller M, Gillquist J, Ekstrand J. Isokinetic torque levels for knee extensors and knee flexors in soccer players. Int J Sports Med. 1986;7:50-53.

[122] Read MT, Bellamy MJ. Comparison of hamstring/quadriceps isokinetic strength ratios and power in tennis, squash, and track athletes. Br J Sports Med. 1990;24:178-182.

[123] Snow CJ, Blacklin K. Reliability of knee flexor peak torque measurements from a standardized test protocol on a Kin/Com dynamometer. Arch Phys Med Rehabil. 1992;73:15-21.

[124] Perrin DH, Robertson RJ, Ray RL. Bilateral isokinetic peak torque, torque acceleration energy, power, and work relationships in athletes and nonathletes. J Orthop Sports Phys Ther. 1987;9:184-189.

[125] Hoffman T, Stauffer RW, Jackson AS. Sex difference in strength. Am J Sports Med. 1979;7:265-267.

[126] Tabin GC, Gregg JR, Bonci T. Predictive leg strength values in immediately prepubescent and postpubescent athletes. Am J Sports Med. 1985;13:387-389.

[127] Callister R, Callister RJ, Staron RS, et al. Physiological characteristics of elite judo athletes. Int J Sports Med. 1991;12:196-203.

[128] Hageman PA, Mason DK, Rydlund KW, Humpal SA. Effects of position and speed on eccentric and concentric isokinetic testing of the shoulder rotators. J Orthop Sports Phys Ther. 1989;11:64-69.

[129] Jackson AS, Pollock ML. Factor analysis and multivariate scaling of anthropometric variables for the assessment of body composition. Med Sci Sports. 1976;8:196-203.

[130] Barr AK, Duncan PW. Influence of position on knee flexor peak torque. J Orthop Sports Phys Ther. 1988;9:279-283.

[131] Bohannon RW, Smith MB. Intrasession reliability of angle specific knee extension torque measurements with gravity correction. J Orthop Sports Phys Ther. 1989;11:155-157.

[132] Burdett R, Swearingen JV. Reliability of isokinetic muscle endurance tests. J Orthop Sports Phys Ther. 1987;8:484-488.

[133] Cawthorn M, Cummings G, Walker JR, Donatelli R. Isokinetic measurement of foot invertor and evertor force in three positions of planter and dorsiflexion. J Orthop Sports Phys Ther. 1991;14:75-81.

[134] Appen L, Duncan PW. Strength relationship of the knee musculature: effects of gravity and sport. J Orthop Sports Phys Ther. 1986;7:232-235.

[135] Magnusson SP, Gleim GW, Nicholas JA. Shoulder weakness in professional baseball pitchers. Med Sci Sports Exerc. 1994;26:5-9.

[136] McMaster WC, Long SC, Caizzo VJ. Shoulder torque changes in the swimming athlete. Am J Sports Med. 1992;20:323-327.

[137] Taylor NA, Cotter JD, Stanley SN, Marshall RN. Functional torque-velocity and power-velocity characteristics of elite athletes. Eur J Appl Physiol. 1991;62:116-121.

[138] Francis K, Hoobler T. Comparison of peak torque values of the knee flexor and extensor muscle groups using the Cybex II and Lido 2.0 isokinetic dynamometers. J Orthop Sports Phys Ther. 1987;8:480-483.

[139] Coyle EF, Costill DL, Lesmes GR. Leg extension power and muscle fiber composition. Med Sci Sports. 1979;11:12-15.

[140] Hanten WP, Ramberg CL. Effect of stabilization on maximal isokinetic torque of the quadriceps femoris muscle during concentric and eccentric contractions. Phys Ther. 1988;68:219-222.

[141] Knapik JJ, Ramos M. Isokinetic and isometric relationships in the human body. Arch Phys Med Rehabil. 1980;61:64-67.

[142] Leslie M, Zachazewski J, Browne P. Reliability of isokinetic torque values for ankle invertors and evertors. J Orthop Sports Phys Ther. 1990;11:612-615.

[143] Otis J, Godbold J. Relationship of isokinetic torque to isometric torque. J Orthop Res. 1983;1:165-171.

[144] Westing SH, Seger JY, Karlson E, Ekblom B. Eccentric and concentric torque-velocity characteristics of the quadriceps femoris in man. Eur J Appl Physiol. 1988;58:100-104.

[145] Worrel TW, Perrin DH, Denegar CR. The influence of hip position on quadriceps and hamstring peak torque and reciprocal muscle group ratio values. J Orthop Sports Phys Ther. 1989;11:104-107.

[146] Molnar GE, Alexander J. Objective, quantitative muscle testing in children: a pilot study. Arch Phys Med Rehabil. 1973;54:224-228.

[147] Gibson JNA, Scott M. Long-term effects of patellectomy on quadriceps and hamstring isokinetic function. Physiotherapy. 1991;77:711-714.

[148] Lorentzon R, Elmqvist LG, Sjostrom M, et al. Thigh musculature in relation to chronic anterior cruciate ligament tear: muscle size, morphology, and mechanical output before reconstruction. Am J Sports Med. 1989;17:423-429.

[149] Stratford P. Reliability of a peak knee extensor and flexor torque protocol: a study of post ACL reconstructed knees. Physiotherapy Canada. 1991;43:27-30.

[150] Pitetti KH, Climstein M, Mays MJ, Barrett PJ. Isokinetic arm and leg strength of adults with Down syndrome: a comparative study. Arch Phys Med Rehabil. 1992;3:847-850.

[151] Moffroid MA, Whipple R, Hofkosh J, et al. A study of isokinetic exercise. Phys Ther. 1969;49:735-746.

[152] Kramer JF, Vaz MD, Hakansson D. Effect of activation force on knee extensor torques. Med Sci Sports Exerc. 1991;23:231-237.

[153] Klopfer DA, Greij SD. Examining quadriceps/hamstrings performance at high velocity isokinetics in untrained subjects. J Orthop Sports Phys Ther. 1988;10:18-22.

[154] Kibler WB, Chandler TJ, Uhl T, Maddux R. A musculosketal approach to the preparticipation physical examination: preventing injury and improving performance. Am J Sports Med. 1989;17:525-531.

[155] Lucca JA, Kline KK. Effects of upper and lower limb preference on torque production in the knee flexors and extensors. J Orthop Sports Phys Ther. 1989;11:202-207.

[156] Herzog W, Guimaraes AC, Anton MG, Carter EK Moment-length relations of rectus femoris muscles of speed skaters/cyclists and runners. Med Sci Sports Exerc. 1991;23:1289-1296.

[157] Harding B, Black T, Bruulsema A, et al. Reliability of a reciprocal test protocol performed on the kinetic communicator: an isokinetic test for knee extensor and flexor strength. J Orthop Sports Phys Ther. 1988;10:218 -223.

[158] van der Leeuw GH, Stam HJ, v Nieuwenhuyzen JF. Correction for gravity in isokinetic dynamometry of knee extensors in below-knee amputees. Scand J Rehabil Med. 1989;21:141-145.

[159] 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.

[160] Kannus P. The relationship between peak torque and work of the quadriceps and hamstrings after knee injury. J Sports Med Phys Fitness. 1990;30:185-189.

[161] Wilhite MR, Cohen ER, Wilhite S. Reliability of concentric and eccentric measurements of quadriceps performance using the Kin-Com dynamometer: the effect of testing order for three different speeds. J Orthop Sports Phys Ther. 1992;15:175-182.

[162] Hart DL, Miller LC, Stauber WT. Effect of cooling on force oscillations during maximal voluntary eccentric exercise. Exp Neurol. 1985;90:73-80.

[163] Hortobagyi T, Katch FI, LaChance PF. Interrelationships among various measures of upper body strength assessed by different contraction modes: evidence for a general strength component. Eur J Appl Physiol 1989;58:749-755.

[164] Kramer JF. Reliability of knee extensor and flexor torques during continuous concentric-eccentric cycles. Arch Phys Med Rehabil. 1990;71: 460-464.

[165] Nunn KD, Mayhew JL. Comparison of three methods of assessing strength imbalances at the knee. J Orthop Sports Phys Ther. 1988;10:134-137.

[166] Perrine JJ, Edgerton VR. Muscle force-velocity and power-velocity relationships under isokinetic loading. Med Sci Sports. 1978;10:159-166.

[167] Epler M, Nawoczenski D, Englehart T. Comparison of the Cybex II standard shin adapter versus the Johnson antishear device in torque production. J Orthop Sports Phys Ther. 1988;9:284-286.

[168] Grabiner MD, Hawthorne DL. Conditions of isokinetic knee flexion that enhance isokinetic knee extension. Med Sci Sports Exerc. 1990;22:235-244.

[169] Grabiner MD, Jeziorowski JJ, Diverkar AD. Isokinetic measurements of trunk extension and flexion performance collected with the Biodex clinical data station. J Orthop Sports Phys Ther. 1990;11:590-598.

[170] Gransberg L, Knutsson E. Determination of dynamic muscular strength in man with acceleration-controlled isokinetic movements. Acta Physiol Scand. 1983;119:317-320.

[171] Jorgensen K. Force-velocity relationship in human elbow flexors and extensors. Biomechanica. 1976;V:145-151.

[172] Kannus P, Jarvinen M. Nonoperative treatment of acute knee ligament injuries: a review with special reference to indications and methods. Sports Med. 1990;9:244-260.

[173] Kannus P,Jarvinen M. Maximal peak torque as a predictor of peak angular impulse and average power of thigh muscles: an isometric and isokinetic study. Int J Sports Med. 1990;11: 146-149.

[174] Hart DL, Stobbe TJ, Till CW, Plummer RW. Effect of trunk stabilization on quadriceps femoris muscle torque. Phys Ther. 1984;64: 1375-1380.

[175] Burnett CN, Betts EF, King WM. Reliability of isokinetic measurements of hip muscle torque in young boys. Phys Ther. 1990;70:244-249.

[176] Housh TJ, Johnson CO, Hughes RA, et al. Isokinetic strength and body composition of high-school wrestlers across age. Med Sci Sports Exerc. 1989;21:105-109.

[177] Gross MT, Credle JK Hopkins LA, Kollins TM. Validity of knee flexion and extension peak torque prediction models. Phys Ther. 1990;70:3-10.

[178] Herzog W, Hasler E, Abrahamse SK A comparison of knee extensor strength curves obtained theoretically and experimentally. Med Sci Sports Exerc. 1991;23:108-114.

[179] Feiring DC, Ellenbecker TS, Derscheid GL. Test-retest reliability of the Biodex Isokinetic Dynamometer. J Orthop Sports Phys Ther. 1990;11: 298-300.

[180] Karnofel H, Wilkeinson K Lentell G. Reliability of isokinetic muscle testing at the ankle. J Orthop Sports Phys Ther. 1989;11:150-154.

[181] Gray JC, Chandler JM. Percent decline in peak torque production during repeated concentric and eccentric contractions of the quadriceps femoris muscle. J Orthop Sports Phys Ther. 1989;10:309-314.

[182] Montgomery LC, Douglass LW, Deuster PA. Reliability of isokinetic test of muscle strength and endurance. J Orthop Sports Phys Ther. 1989;11:315-322.

[183] Mawdsley RH, Croft BJ. The effects of submaximal contractions on an isokinetic test session. J Orthop Sports Phys Ther. 1982;4:74.

[184] Kues JM, Rothstein JM, Lamb RL. Obtaining reliable measurements of knee extensor torque produced during maximal voluntary contractions: an experimental investigation. Phys Ther. 1992;72:492-504.

[185] Balogun JA, Akomolafe CT, Amusa LO. Grip strength: effects of testing posture and elbow position. Arch Phys Med Rehabil. 1991;72:280-283.

[186] Worrell TW, Denegar CR, Armstrong SL, Perrin DH. Effect of body position on hamstring muscle group average torque. J Orthop Sports Phys Ther. 1990;11:449-452.

[187] Komi PV. Measurement of the force-velocity relationship in human muscle under concentric and eccentric contractions. Medicine and Sport. 1973;8:224-229

[188] Patterson ME, Nelson SG, Duncan PW. Effects of stabilizing the nontested lower extremity during isokinetic evaluation of the quadriceps and hamstrings. J Orthop Sports Phys Ther. 1984;6:18-20.

[189] Stokes IA, Gookin DM, Reid S, Hazard RG. Effects of axis placement on measurement of isokinetic flexion and extension torque in the lumbar spine. J Spinal Disord. 1990;3:114-118.

[190] Jensen RC, Warren B, Laursen C, Morrissey MC. Static pre-load effect on knee extensor isokinetic concentric and eccentric performance. Med Sa Sports Exerc. 1991;23:10-14.

[191] Lord JP, Aitkens SG, McCrory MA, Bernauer EM. Isometric and isokinetic measurement of hamstring and quadriceps strength. Arch Phys Med Rehabil. 1992;73:324-330.

[192] Otis JC, Gould J. The effect of external load on torque production by knee extensors. J Bone Joint Surg [Am]. 1986;68:65-70.

[193] Friedlander AL, Block JE, Byl NN, et al. Isokinetic limb and trunk muscle performance testing: short-term reliability. J Orthop Sports Phys Ther. 1991;14:220-224.

[194] Gross MT, McGrain P, Demilio N, Plyler L. Relationship between multiple predictor variables and normal knee torque production. Phys Ther. 1989;69:54-62.

[195] Osternig LR. Optimal isokinetic loads and velocities producing muscular power in human subjects. Arch Phys Med Rehabil. 1975;56:152-155.

[196] Kramer JF, Hill K, Jones IC, et al. Effect of dynamometer application arm length on concentric and eccentric torques during isokinetic knee extension. Physiotherapy Canada. 1989;41:100-105.

[197] Taylor RL, Casey JJ. Quadriceps torque production on the Cybex II dynamometer as related to changes in lever arm length. J Orthop Sports Phys Ther. 1986;8:147-152.

[198] Gravel D, Richards CL, Filion M. Influence of contractile tension development on dynamic strength measurements of the plantarflexors in man. J Biomech. 1988;21:89-96.

[199] Piette V, Richards C, Filton M. Use of static pre-loading in estimation of dynamic strength with the Kin-Com dynamometer. Presented at the North American Congress on Biomechanics; 1986; Montreal, Quebec, Canada.

[200] Wessel J, Gray G, Luongo F, Isherwood L. Reliability of work measurements recorded during concentric and eccentric contractions of the knee extensors in healthy subjects. Physiotherapy Canada. 1989; 41:250-253.

[201] Tredinnick TJ, Duncan PW. Reliability of measurements of concentric and eccentric isokinetic loading. Phys Ther. 1988;68:656-659.

[202] Sapega AA, Nicholas JA, Sokolow D, Saraniti A. The nature of torque "overshoot" in Cybex isokinetic dynamometry. Med Sci Sports Exerc. 1982;14:368-375.

[203] Rothstein JM, Delitto A, Sinacore DR, Rose SJ. Muscle function in rheumatic disease patients treated with corticosteroids. Muscle Nerve. 1983;6:128.

[204] Rathfon 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.

[205] Winter DA, Wells RP, Orr GW. Errors in the use of isokinetic dynamometers. Eur J Appl Physiol. 1981;46:397-408.

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

[207] Dudley CA, Harris RT, Duvoisin MR, et al. Effect of voluntary vs artificial activation on the relationship of muscle torque to speed. J Appl Physiol. 1990;69:2215-2221.

[208] Ghena DR, Kurth AL, Thomas M, Mayhew J. Torque characteristics of the quadriceps and hamstring muscles during concentric and eccentric loading. J Orthop Sports Phys Ther. 1991;14:149-154.

[209] Hartsell HD, Kramer JF. A comparison of the effects of electrode placement, muscle tension, and isometric torque of the knee extensors. J Orthop Sports Phys Ther. 1992;15:168-174.

[210] Levene JA, Hart BA, Seeds RH, Fuhrman GA. Reliability of reciprocal isokinetic testing of the knee extensors and flexors. J Orthop Sports Phys Ther. 1991; 14:121-127.

[211] Molczyk L, Thigpen LK, Eickhoff J, et al. Reliability of testing the knee extensors and flexors in healthy adult women using a Cybex II isokinetic dynamometer. J Orthop Sports Phys Ther. 1991;14:37-41.

[212] Pitetti KH. A reliable isokinetic strength test for arm and leg musculature for mildly mentally retarded adults. Arch Phys Med Rehabil. 1990;71:669-672.

[213] Cress ME, Johnson J, Agre JC. Isokinetic strength testing in older women: a comparison of two systems. J Orthop Sports Phys Ther. 1991;13:199-202.

[214] Nelson SG, Duncan PW. Correction of isokinetic and isometric torque recordings for the effects of gravity. Phys Ther. 1983;63:674-676.

[215] Hortobagyi T, Katch FI. Role of concentric force in limiting improvement in muscular strength. J Appl Physiol. 1990;68:650-658.

[216] Walmsley RP, Szybbo C. A comparative study of the torque generated by the shoulder internal and external rotator muscles in different positions and at varying speeds. J Orthop Sports Phys Ther. 1987;9:217-222.

[217] 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: isometric and isokinetic evaluation. Am J Sports Med. 1992;20:162-168.

[218] Kannus P, Jarvinen M. Thigh muscle function after partial tear of the medial ligament compartment of the knee. Med Sci Sports Exerc. 1991;23:4-9.

[219] Stratford PW, Bruulsema A, Maxwell B, et al. The effect of inter-trial rest interval in the assessment of isokinetic thigh muscle torque. J Orthop .Sports Phys Ther. 1990;11:362-366.

[220] Patton RW, Hinson MM, Arnold BR, Lessard B. Fatigue curves of isokinetic contractions. Arch Phys Meld Rehabil 1978;59:507-509.

[221] Baltzopoulos V, Williams JG, Brodie DE. Sources of error in isokinetic dynamometry: effects of visual feedback on maximum torque measurements. J Orthop Sports Phys Ther. 1991;13:138-142.

[222] Figoni SF, Morris AF. Effects of knowledge of results on reciprocal, isokinetic strength and fatigue. J Orthop Sports Phys Ther. 1984;6:190-197.

[223] Peacock B, Westers S, Walsh S, Nicholson K Feedback and maximum voluntary contraction. Ergonomics. 1981;24:223-228.

[224] Estlander AM, Mellin G, Weckstrom A. Influence of repeated measurements on isokinetic lifting strength. Clin Biomech. 1992;7:149-152.

JL Keating, PT, is pursuing a doctorate by research thesis in the Department of Behavioural Health Sciences, Lincoln School of Health Sciences, La Trobe University, Bundoora, Victoria, Australia 3083 ([e-mail]) Address all correspondence to Ms Keating.

TA Matyas, PhD, is Reader, Department of Behavioural Health Sciences, and Convenor, Movement Rehabilitation Research Group, Lincoln School of Health Sciences, La Trobe University.

This work was funded by the Harold Hopkins Award from the Hospital Superannuation Board, Victoria, Australia; the Felice Rosemary Lloyd scholarship; and the Australian Post Graduate Research Award.

This article was submitted November 20, 1995, and was accepted March 20, 1996.