Reliability and validity of shoulder function outcome measures in people with a proximal humeral fracture

Abstract

Purpose: Investigate test--retest reliability and validity of five shoulder outcome measures in people during their active rehabilitation after a shoulder fracture. Method: This prospective longitudinal study assessed shoulder function in 20 people (16 women, mean age 68.1 years) with surgical or conservative management at 6, 12 and 13 weeks post proximal humeral fracture using three patient-reported (Disabilities of Arm, Shoulder and Hand; Oxford Shoulder Score; Subjective Shoulder Value) and two clinician-administered (Constant Score; UCLA Shoulder score) outcome measures. Results: Content analysis categorised items into multiple domains of functioning for each outcome measure. Construct validity testing between measures found moderate to strong correlations (r?=?0.43–0.92). Longitudinal validity (responsiveness), represented by correlations between change scores, was moderate to strong (r?=?0.44–0.83). Although ICCs_2,1 for test–retest reliability ranged from 0.75 to 0.93, Limits of Agreement between measurements were relatively wide (10–23% of available range of scores). Minimal clinically important difference estimates varied between anchor- and distribution-based methods. Conclusions: The five outcome measures assessing shoulder function provided values for reliability and validity that meet measurement requirements for use in groups of people after a proximal humeral fracture. However, the use of these outcome measures might be limited by low absolute agreement between measurements and their content covering multiple domains of functioning.

• Implications for Rehabilitation

• Linking the International Classification of Functioning, Disability and Health (ICF) to the content of common shoulder function outcome measures showed that multiple domains of functioning are combined into a single score. This might not be preferred for measurement of the single construct of “shoulder function”.

• Currently available shoulder function outcome measures may not be sufficiently reliable to monitor change in an individual after a proximal humeral fracture during the rehabilitation phase.