Keele Research Repository

Background
Patients with musculoskeletal pain in different body sites share common prognostic factors. Using prognosis to stratify and treatment match can be clinically and cost-effective. We aimed to refine and validate the Keele STarT MSK tool for prognostic stratification of musculoskeletal pain patients.

Methods
Tool refinement and validity was tested in a prospective cohort study, and external validity examined in a pilot cluster RCT. Study population comprised 2414 adults visiting UK primary care with back, neck, knee, shoulder or multisite pain returning postal questionnaires (cohort: 1890 (40% response); trial: 524). Cohort baseline questionnaires included a draft tool plus refinement items. Trial baseline questionnaires included the Keele STarT MSK tool. Physical health (SF-36 Physical Component Score (PCS)) and pain intensity were assessed at 2- and 6-months cohort follow-up; pain intensity was measured at 6-months trial follow-up.

Results
The tool was refined by replacing (3), adding (3) and removing (2) items, resulting in a 10-item tool. Model fit (R2) was 0.422 and 0.430 and discrimination (c-statistic) 0.839 and 0.822 for predicting 6-month cohort PCS and pain (respectively). The tool classified 24.9% of cohort participants at low, 41.7% medium and 33.4% high risk, clearly discriminating between subgroups. The tool demonstrated model fit of 0.224 and discrimination 0.73 in trial participants. Multiple imputation confirmed robustness of findings.

Conclusions
The Keele STarT MSK tool demonstrates good validity and acceptable predictive performance, and clearly identifies groups of musculoskeletal pain patients with different characteristics and prognosis. Using prognostic information for stratification and treatment matching may be clinically/cost-effective.