Reliability of Different Dynamometers Used to Measure Isometric Shoulder Strength: A Comparative Study of Two Dynamometers in 526 Adults.

Background: Several devices are available to record isometric shoulder strength; however, minimal information regarding inter-device reliability and agreement is available. The objective of this study was to determine inter-device reliability and agreement of two dynamometers for measuring shoulder abduction strength. Methods:Five hundred and twenty-six adults had their bilateral isometric 90o shoulder abduction strength recorded with both the ChatillonTM force gauge and IDOTM Isometer devices by 10 independent raters. Intra-class correlation coefficients (ICCs), 95% limits of agreement (95% LoA), and minimum detectable change were calculated to determine inter-device reliability and agreement. Inter-rater and intra-rater reliability was measured in a small cohort to exclude this being a factor reducing validity. Results:Inter-device reliability was excellent for both devices with an ICC of 0.89 (0.87-0.91). There is a significant positive bias with the ChatillonTM force gauge measuring a mean of 0.62 kg (95% confidence interval (CI) 0.47 - 0.77 kg) more than the IDOTM device. Inter-device limits of agreement were found to be approximately 50% of mean strength measurement, being dependent on the strength of the individual, with increasing inter-device error as strength increases. Inter-rater reliability was excellent for both devices, with ChatillonTM achieving a superior ICC and a narrower 95% CI with 0.92 (0.80-0.97) compared to IDOTM with 0.85 (0.05-0.96). Intra-rater reliability was excellent with both the ChatillonTM and IDOTM devices, with an ICC of 0.96 (95% CI 0.91 - 0.99) and 0.95 (95% CI 0.88 - 0.98), respectively. Conclusions: With high inter-device reliability, either individual device is suitable for measuring and comparing shoulder abduction strength longitudinally. However, consistency in assessors and device use is recommended, as device measurement error may be statistically significant and may exceed minimal clinical differences for certain scores. As a consequence, care must be taken when comparing published results on shoulder abduction strength, due to variation not only in device measurement error, but also in study protocols in general.