Determining the Interrater Reliability of the SOFMER Activity Score (version 2) for Individuals in Rehabilitation Centers

Abstract

ObjectiveTo assess the interrater reliability of the SOFMER Activity Score (SAS) (version 2 [v2], an 8-item [4 motor and 4 cognitive] and 5-level scale) and improve its scoring system before conducting further validation steps. DesignCross-sectional, prospective, observational, noninterventional, and multicentric study. SettingThe study was conducted between November 2018 and September 2019 in 4 French rehabilitation centers (2 public university hospitals for adults and 2 private not-for-profit rehabilitation centers for children). ParticipantsThe study included 101 participants (N=101; mean age, 44.5±25.4 years; 28.7% younger than 18 and 18.8% older than 65 years). The female/male sex ratio was 0.6. The causes for admission to the center were mainly neurologic (65%) or orthopedic (24%). InterventionsNone. Main Outcome MeasuresActivity limitation was rated with the SAS the same day by 2 independent multidisciplinary teams. The interrater reliabilities of the score items were assessed using weighted kappa coefficients. ResultsAll weighted kappa coefficients ranged between 0.83 and 0.92, indicating “good” to “excellent” interrater reliability. Interteam score disagreements occurred in 227 of 808 scores (28%). The reason for most disagreements was unnoticed human or material aid during the observation period. ConclusionsThe results demonstrate the high interrater reliability of the SASv2 and allow carrying out further validation steps after minor changes to item scoring instructions and clearer definitions of some items that help improving scoring standardization. The SASv2 may then become a consistent measure of activity level for clinical research or burden of care investigations.