Development and validation of a short version of the French Hand Function Sort questionnaire in vocational rehabilitation.

The Hand Function Sort (HFS) is a pictorial self-administered questionnaire with 62 items. It is a valid and reliable scale focused on the physical function of the upper limbs. It is used to predict the return to work. We aimed to develop and validate a short version of the French version of the HFS (HFS-F) to simplify its use in clinical practice. We included patients with upper-limb chronic pain hospitalised for vocational rehabilitation from 2012 to 2019. Vocational rehabilitation aims to improve the autonomy of patients to regain their previous working capacity. The 62 items of the HFS-F were analysed in terms of patient and expert assessments, floor/ceiling effect, item-to-total correlation, principal component analysis, and Rasch analysis. A short HFS-F was developed. Thereafter, we assessed its internal consistency, test-retest reliability, criterion validity with the full-length HFS-F, construct validity with different scales (Disabilities of the Arm, Shoulder, and Hand [DASH]; Brief Pain Inventory [BPI]; Hospital Anxiety and Depression [HAD]), standard error of measurement (SEM), and minimal detectable change (MDC). Six experts were consulted, 34 patients were interviewed, and 629 questionnaires were analysed. Among the items, 25 were selected after the final round with the six experts. The internal consistency and test-retest reliability were excellent (Cronbach α=0.95, intraclass correlation coefficient=0.92, 95% confidence interval [95% CI] 0.87 to 0.95). The correlation coefficient between scores of the short and full-length HFS-F was 0.841 (95% CI: 0.752 to 0.897, P<10-4), and those between the short HFS-F score and the DASH, BPI, HAD-Anxiety, and HAD-Depression scores were -0.816 (95% CI: -0.714 to -0.881, P<10-4), -0.529 (95% CI: -0.338 to -0.674, P<10-4), -0.451 (95% CI: -0.244 to 0.614, P=0.0001), and -0.360 (95% CI: -0.140 to -0.542, P=0.0018), respectively. The SEM and MDC values were estimated at 6/100 and 17/100, respectively. A short version of the HFS-F was developed and validated. We named this questionnaire the 25 HFS-F