Subject variability of shoulder abduction strength testing.

To quantify normal biological variability of shoulder abduction strength testing with an isokinetic and a hand-held dynamometer, we tested nine healthy subjects over a clinically relevant period. One side was tested with a Cybex at 60 deg/sec and the other with a hand-held dynamometer. Six maximal trials, following warmup, were conducted on each of 5 days, separated by 1 to 2 weeks. Intraday correlations of individual trials ranged from 0.82 to 0.995 for hand-held dynamometer, and 0.88 to 0.996 for Cybex. Interday correlations ranged from 0.94 to 0.98 for hand-held dynamometer, and 0.88 to 0.97 for Cybex. The best values to use for Cybex interday variability were the average of the first three repetitions, and yielded standard errors of 8.6% to 19.2% of the sample mean. The average of the last three repetitions were the best for hand-held dynamometers, and yielded standard errors of 5.5% to 10.8%. There was a significant decline (P less than 0.05) in strength of the mean of the last three versus the mean of the first three daily repetitions on Day 1 and 4 for hand-held dynamometers, and a trend toward this on the other days, that was not seen with Cybex. Regression of average Day 1 values for Cybex and hand-held dynamometer yielded r = 0.86 (P less than 0.01) with a slope of 1.07 indicating good agreement between modalities. In conclusion, intraday/interday correlations were high for Cybex and hand-held dynamometer. Interday variability was minimized by using the mean score of the first three and last three repetitions for Cybex and hand-held dynamometer, respectively. Changes in Cybex and hand-held dynamometer strength of less than 19% and 11%, respectively, are within the area of "measurement error" and should therefore not be considered clinically significant.