Stratifying workers on sick leave due to musculoskeletal pain: translation, cross-cultural adaptation and construct validity of the Norwegian Keele STarT MSK tool

Stratified care using prognostic models to estimate the risk profiles of patients has been increasing. A refined version of the popular STarT Back tool, the Keele STarT MSK tool, is a newly developed model for matched treatment across a wide range of musculoskeletal pain presentations. The aim of this study was to translate and culturally adapt the Keele STarT MSK tool into Norwegian, examine its construct validity and assess the representativeness of the included sample. The Keele STarT MSK tool was formally translated into Norwegian following a multistep approach of forward and backward translation. A pre-final version was tested in 42 patients. Minor changes were implemented. To assess its construct validity, an online survey was conducted among workers aged 18-67 years who were on sick leave (>4 weeks) due to musculoskeletal disorders. Construct validity was evaluated in terms of convergent and discriminant validity using Pearson's correlation coefficient, and known-group validity by comparing risk subgroups as suggested by the COSMIN checklist. The representativeness of the sample was assessed by comparing demographic and sick leave information of participants to eligible non-participants (n=168,137). A representative sample of 549 workers participated in the validity assessment; 74 participants (13.5%) were categorised as low risk, 314 (57.2%) as medium risk and 161 (29.3%) as high risk. The construct validity was found sufficient, with 90.9% and 75.0% of the pre-defined hypotheses confirmed for convergent and discriminant validity, and known-group validity, respectively. Floor or ceiling effects were not found. The Keele STarT MSK tool was successfully translated into Norwegian. The construct validity of the tool was acceptable in a representative cohort of workers on sick leave as a result of musculoskeletal pain. However, the analyses raised concerns as to whether one of the questions captures the construct it is intended to measure.