• A Keller, J Hellesnes, and J I Brox.
• Department of Physical Medicine and Rehabilitation, University Hospital Ullevaal, Oslo, Norway. anne.keller@ioks.uio.no
• Spine. 2001 Apr 1;26(7):771-7.

Study DesignA reliability study was performed.ObjectiveTo determine the intersession reproducibility of the isokinetic trunk extensor strength test, the Biering-Sørensen test, and the Astrand test by calculation of the intraclass correlation coefficient and the critical difference.Summary Of Background DataSeveral physical measurements have been used in clinical evaluation of patients with chronic low back pain. Reliability can be evaluated by calculation of either intraclass correlation coefficient or critical difference, but interpretation of the results may vary depending on which method is used.MethodsFor this study, 31 patients with chronic low back pain carried out the isokinetic trunk extensor test, the Biering-Sørensen test, and the Astrand test as also did age- and gender-matched healthy individuals. Measurements were performed in three separate sessions at the same time of the day and by the same experienced examiner. The intervals between the sessions were 5 to 10 days.ResultsThe isokinetic tests showed a learning effect between Tests 1 and 2. For the isokinetic test, the critical difference increased with increased angular velocities from 28% to 63% in the patients and from 27% to 39% in the healthy subjects. The critical differences for the patients and the healthy individuals were 57% vs 54% for the Biering-Sørensen test and 21% vs 23% for the Astrand test, respectively. The intraclass correlation coefficient ranged from 93% to 98% in the patients and 80%-98% in the healthy individuals.ConclusionsThe reliability was found to be acceptable for the Astrand test and the isokinetic test at 60 degrees per second, as evaluated by the critical difference. All the tests were highly reliable according to the intraclass correlation coefficient, except for Biering-Sørensen test for healthy individuals. The critical difference is the preferable measure because calculation of the intraclass correlation coefficient may give a misleading high estimate of reliability.

18 keywords...

hide…