Generalizability of a Musculoskeletal Therapist Electronic Health Record for Modelling Outcomes to Work-Related Musculoskeletal Disorders.

The data came from one organisation and one electronic health record system, so they may not represent other clinics, professions or systems. The records were overwhelmingly from manufacturing (95.4%), so the dataset does not show how other industries behave. Duration of employment had 10.6% missing and 17.5% implausible records, and nationality data were affected by default settings and clinicians recording cultural background in the nationality field. The comparison datasets had different purposes and coding categories, especially for mechanism of injury and diagnosis, so direct matching was incomplete. Analysis was at the disorder event level, and one worker could have multiple records, so workers with repeated injuries may be overrepresented. The study did not assess the overall data quality or validate a prediction model, so it is a first step rather than proof that models built from the data will work. Many potentially important variables, such as psychosocial factors, job demands and workplace modifications, were not analysed because comparable national data were unavailable.

Royal Melbourne Institute of Technology is named as funder; no other conflict statement is supplied.

Do not read this as evidence that onsite musculoskeletal therapy improves work-related injury outcomes. The paper only describes an electronic health record dataset and compares its demographics and injury categories to national workforce and claims data; it did not test a treatment or build a validated prediction model.