This study tested knee extensor and flexor strength using a reciprocal testing protocol on the Kin-Corn@ isokinetic dynamometer. The purposes were: 1) to identify the variability associated with subjects, repetitions, occasions, and their various interactions; 2) to express the overall measurement error in clinically relevant terms; and 3) advance a strategy for maximizing reliability while simultaneously ensuring that all major sources of measurement error are represented. Fourteen healthy women were tested on two separate occasions (test sessions). On each occasion, measurements of knee extensor and flexor peak torque, average torque, and peak torque angle were gathered for each subject. The results indicated a high overall reliability (range 0.936-0.952) for all measures except peak torque angle (0.630-0.799). Reliability coefficients were consistently lower between occasions than among repetitions, indicating that this type of testing should occur on more than one occasion so as to include all sources of measurement error. This study has expressed measurement error in clinically relevant terms using a 95% confidence range. The findings indicate that a high reliability for isokinetic leg strength measurements can be achieved using a reciprocal testing protocol on the Kin-Com device.lsokinetic exercise and strength testing devices analysis should only include the last three repetiare frequently used by clinicians and researchers t i o n~.~ When evaluating this conclusion, it is imas assessment tools. In order for a measurement portant that the reader keep in mind that peak system to be considered clinically useful, there torque among the six repetitions varied by no must be evidence that the device is reliable, valid, rr10t-e than 2.05 fi-lb, suggesting that, although and sensitive to clinically important change.' The statistically significant, the results may not be focus of this paper was reliability as it is the first clinically important-Mawdsley and Knapik5 in a prerequisite for validity and sensitivity to change. study of knee extension performed on 16 healthy Johnson and Siegel,4 in a study of isokinetic volunteers at 3O0/sec, demonstrated different knee extension performed at 18001sec on 40 peak torque repetition patterns when measured females, reported an intraclass correlation coeffi-on different test sessions. A reliability coefficient cient of omg7 based on six repetitions on each of was not cited by these authors. White and Protwo test sessions. A modest linear trend in peak tas," reporting on a study of knee extension torque was demonstrated over repetitions and as performed at 60°/sec on 30 females, demonstrated statistically significant differences in peak this trend was principall~ accounted for the torque among occasions, but not between repefirst three repetitions, Johnson suggested that titions. White and Protas further reported intraclass correlation coefficients of 0.79 for a single 'Clinical measurement consultant, Hamilton, Ont. repetition and 0.96 based on the mean of the last t Physiot...
