Can risk stratification be applied to common musculoskeletal conditions? Refinement and validation of the Keele STarT MSK tool

Talk Code: 
Paul Campbell
Kate M. Dunn, Ebenezer K. Afolabi, Martyn Lewis, Danielle van der Windt, Jonathan C. Hill, Christian D. Mallen, Joanne Protheroe, Elaine M. Hay, Nadine E. Foster
Author institutions: 
Keele University


Patients with musculoskeletal pain in different body regions share common prognostic factors. Back pain research demonstrates that using prognostic information to subgroup patients and match to treatment pathways can be clinically and cost-effective. Similar approaches could also benefit patients with musculoskeletal pain in other body regions. Initial analyses highlighted that modifying a back pain stratification tool (STarT Back tool) was promising, but the tool required refinement for use among patients with musculoskeletal pain in other body regions. We aimed to refine and validate an instrument, the Keele STarT MSK tool, for sub-grouping patients consulting in primary care with the five most common musculoskeletal pain presentations.


Patients consulting with musculoskeletal pain in the back, neck, knee, shoulder or at multiple body sites, at one of 14 general practices, were invited to participate in a prospective cohort study. Postal data were collected at baseline, two- and six-months. Questionnaires included the modified STarT Back tool, physical health (SF-36 Physical Component Score (PCS)) and pain intensity, plus candidate items for refining the tool. Refinement aimed to maximise prognostic performance (prediction and discrimination at two-months). Optimal subgroup cut-points were chosen based on a-priori defined levels of sensitivity, specificity, predictive values and likelihood ratios to discriminate poor outcome (SF-36 PCS below 37.17) at two-months. Performance across pain sites was considered. Six-month data were used to assess validity.


1890 of 4720 patients responded and consented to the survey. Response at two- and six-months was 76% and 79% respectively. A 9-item refined STarT MSK tool (including three item substitutions) improved performance against six-months outcomes; prediction (goodness of model fit) from an R2 of 0.334 for the draft tool to 0.389 for the refined Tool, and discrimination (AUC) from 0.80 (95% CI 0.78, 0.83) to 0.82 (0.79, 0.84). Subgroup cut-points were 0-3 for low risk (30% of sample), 4-7 for medium risk (51%) and 8-9 for high risk (19%) based on an overall score from 0-9. Subgroups differed significantly in terms of baseline characteristics including pain intensity, mental health, sleep disturbance, health literacy, and work absence. Subgroups also differed in outcomes; 91% of the high risk group, 63% of medium risk and 18% of low risk had poor outcomes at six-months. Performance, cut-points and characteristics and outcomes of subgroups were comparable across pain sites.


The Keele STarT MSK tool is a valid instrument for stratifying patients with back, neck, knee, shoulder or multi-site pain into prognostic subgroups. These subgroups can be matched to appropriate treatment options. A randomised trial is currently testing whether stratified care, combining use of the new tool with matched treatment options, is more effective than usual primary care for patients with the five most common musculoskeletal pain presentations.

Submitted by: 
Paul Campbell
Funding acknowledgement: 
This submission presents independent research funded by the National Institute for Health Research (NIHR) under its Programme Grants for Applied Research scheme (grant number: RP-PG-1211-20010) and the Arthritis Research UK Centre in Primary Care grant (Grant Number 18139). NF and JH are supported through an NIHR Research Professorship awarded to NF (NIHR-RP-011-015). The views expressed in this publication are those of the authors and not necessarily those of the NHS, the NIHR or the Department of Health.