The Shoulder Function Index (SFInX): Evaluation of its measurement properties in people recovering from a proximal humeral fracture

Background: Concerns about test administration, reliability estimations, content and internal structure (dimensionality) of available shoulder measures for people with proximal humeral facture led to the development of a new clinician-observed outcome measure: the Shoulder Function Index (SFInX). The SFInX measures shoulder function by judgement of actual ability to perform daily tasks in which the shoulder is involved. Patients and health professionals had input into the instrument development, and Rasch analysis was used to create a unidimensional, interval-level scale. This study comprehensively evaluated the measurement properties of the SFInX in people recovering from a proximal humeral fracture. Methods: Data were collected on 92 people [79 women, mean age 63.5 years (SD13.9)] who sustained a proximal humeral fracture within the previous year on three occasions to allow for evaluation of the following measurement properties: construct validity (convergent, discriminant and known-groups validity), longitudinal validity (responsiveness), intra-rater reliability (one week retest interval), and inter-rater reliability (n = 20 subgroup; two independent raters). Comparative measures were Constant Score and Disabilities of the Arm Shoulder and Hand (DASH) and discriminative measure was a mental status questionnaire. Minimal clinically important difference, floor and ceiling effects and feasibility of the SFInX were also evaluated. A priori hypotheses were formulated where applicable. Results: Results for construct validity testing supported hypothesised relationships (convergent validity r = 0.75-0.89 (Constant Score and DASH); discriminant validity r = -0.08 (mental status); known-groups validity r = 0.50). For longitudinal validity, lower correlations (r = 0.40-0.49) than hypothesised (r = 0.50-0.70) were found. The SFInX scores changed more (10.3 points) than other scales, which could indicate that the SFInX is more responsive than the comparative measures. Intra-rater and inter-rater reliability found ICCs of 0.96 (95 % CI 0.94-0.97) and 0.91 (95 % CI 0.63-0.97) respectively, with low measurement error (SEM = 3.9-5.8/100). A change of 11-12 points (out of 100) was indicative of a clinically important difference. Conclusions: The SFInX is a feasible outcome measure which clinicians can use to reliably measure and detect clinically important changes in the construct of 'shoulder function', the ability to perform activities in which the shoulder is involved, in people recovering from a proximal humeral fracture.