Reliability by surgical status of self-reported outcomes in patients who have shoulder pathologies

STUDY DESIGN: A test-retest design was used to evaluate the reliability of the self-report sections of 4 shoulder pain and disability scales. OBJECTIVE: The objective of the study was to compare interitem consistency and test-retest reliability by surgical status (postoperative versus nonoperative) and to evaluate the effect of surgical status in the prediction of retest scores. BACKGROUND: Patients and healthcare providers evaluate shoulder status based on self-evaluations of pain and disability. Shoulder outcome measures have been developed that include self-reports, but the properties of these measures have not been assessed by surgical status. METHODS AND MEASURES: A questionnaire containing self-report sections of 4 shoulder scales was administered to study participants twice with 1 week between administrations. The outcome measures examined were the: (1) University of California at Los Angeles (UCLA) Shoulder Score; (2) Constant-Murley Scale (CMS); (3) American Shoulder and Elbow Society (ASES) Shoulder Index; and (4) Shoulder Pain and Disability Index (SPADI). Intraclass correlation coefficients (ICC) were calculated to estimate the test-retest reliability of each of the scales and subscales. The interitem consistencies of the multi-item subscales were assessed using Cronbach's alpha. The effect of surgical status on shoulder outcome scale reliability was evaluated using a general linear models approach. RESULTS: The interitem consistency estimates for the multi-item scales were high with both operative and nonoperative participants (0.88 to 0.96). With the exception of the satisfaction subscale of the UCLA Shoulder Score for the nonsurgical group, the estimated intraclass coefficients ranged from 0.51 to 0.91. The prediction of UCLA-satisfaction and ASES-disability, pain, and total retest scores was improved with the addition of surgical status into a regression model. CONCLUSIONS: The examined scales exhibited good internal consistency across surgical status. The postsurgical sample's reproducibility estimates tended to be higher than those of the nonsurgical sample. Reliability of shoulder outcome scales can be affected by patient surgical status.