Reliability of assessing end-feel and pain and resistance sequence in subjects with painful shoulders and knees

Descriptive. Examine the intrarater and interrater reliability of end-feel and pain/resistance sequence for patients with painful shoulders and knees. Clinicians make diagnostic and intervention decisions based on end-feel and pain/resistance sequence, but few studies have examined agreement within and between physical therapists when assessing subjects with pathology. Subjects with unilateral knee pain (18 men and 22 women with a mean age of 31.8 +/- 9.5 years) or shoulder pain (21 men and 25 women with a mean age of 34.3 +/- 12.9 years) were examined twice. Two physical therapists used standardized positions to evaluate 2 knee motions and 5 shoulder motions. Evaluators did not interview subjects and were blinded to previous test results. Evaluators applied overpressure and noted the end-feel while subjects identified the moment their pain was reproduced. Following testing, subjects rated their pain intensity. Analyses included: percentage of agreement; kappa, weighted kappa, and maximum kappa coefficients; and confidence intervals. Analyses were repeated for subjects whose pain intensity during testing did not change between examinations. Intrarater kappa coefficients varied from 0.65 to 1.00 for end-feel, and intrarater weighted kappa coefficients varied from 0.59 to 0.87 for pain/resistance sequence. Most coefficients remained stable or improved for the unchanged subjects. Interrater kappa coefficients for end-feel and weighted kappa coefficients for pain/resistance sequence varied from -0.01 to 0.70. End-feel kappa coefficients remained low for the unchanged subjects, but pain/resistance sequence weighted kappa coefficients improved. Unbalanced distribution affected many coefficients, producing low coefficients even when the percentage of agreement was high. The appropriate use of end-feel and pain/resistance sequence data requires reliable data gathering, especially when patients are managed by more than one physical therapist. Intrarater reliability of end-feel and pain/resistance judgments at the knee and shoulder were generally good, especially after accounting for subject change and unbalanced distributions. Interrater reliability, however, was generally not acceptable, even after accounting for these factors.