Comparative evaluation of the measurement properties of various shoulder outcome instruments.

BACKGROUND

Various shoulder outcome instruments have been used despite lack of information on their measurement properties; reliability, responsiveness, and validity; and correlation with health-related quality of life.

HYPOTHESIS

Most shoulder outcome instruments have poor correlation with Short Form-36, a general measure of health-related quality of life, and with each other.

STUDY DESIGN

Cohort study (diagnosis); Level of evidence, 2.

METHODS

A consecutive group of 285 patients who had undergone shoulder surgery completed several shoulder outcome instruments-Short Form-36; University of California, Los Angeles shoulder score; American Shoulder and Elbow Surgeons shoulder evaluation form; Constant score; Simple Shoulder Test; Western Ontario Shoulder Instability Index; and the rating sheet for Bankart repair (Rowe score)-preoperatively and at 3, 6, 9, and 12 months postoperatively. Internal consistency, standardized response mean, effect size, and Pearson correlation were used to evaluate reliability, responsiveness, and validity.

RESULTS

The American Shoulder and Elbow Surgeons form, Simple Shoulder Test, and Western Ontario Shoulder Instability Index displayed good internal consistency. The University of California, Los Angeles shoulder score and American Shoulder and Elbow Surgeons form exhibited good responsiveness, whereas Short Form-36 showed the least responsiveness. Pearson correlation coefficients between the shoulder outcome instruments and Short Form-36 were less than excellent (r < .60). Pearson correlation coefficients between the outcome instruments were generally low except for the Constant score and University of California, Los Angeles shoulder score (r = .673, P < .01).

CONCLUSION

There was no single shoulder outcome instrument that was superior to the others in terms of the measurement properties. Most of the tested shoulder outcome instruments did not reflect health-related quality of life well and poorly correlated with each other. This meant that the comparison of a given surgical result with different outcome instruments might be of little practical utility. Further prospective and serial studies should be conducted to develop better shoulder outcome instruments that have significant reliability, responsiveness, validity, and correlation with health-related quality of life. A careful combination of outcome instruments might be necessary to compensate the current evaluation systems.