Prognostic Factors and Treatment Effect Modifiers for Physical Health, Opioid Prescription, and Health Care Utilization in Patients With Musculoskeletal Disorders in Primary Care: Exploratory Secondary Analysis of the STEMS Randomized Trial of Direct Access to Physical Therapist-Led Care.

This was an exploratory secondary analysis of a pilot cluster randomized trial, not a study powered to find which patients benefit most from the direct-access offer. The trial included only 4 general practices in the United Kingdom, and patients were recruited from June 2013 to January 2014, so current practice patterns may differ. The intervention was an offer of direct access, not all participants receiving physical therapy; only 36% in the offer arm accessed physical therapy, and only 33% did so directly. Of 978 participants in the trial, 767 completed 6-month questionnaires and only 457 had complete observations for all variables, so missing data may affect the estimates. Health care utilization was self-reported and combined different investigations and procedures, so it cannot distinguish appropriate from inappropriate care or separate the effects of imaging, injections, and surgery. Prior opioid use was not available, and opioid prescription was measured over 6 months at 4 sites, which may not reflect longer-term prescribing elsewhere. The prognostic models used backward stepwise selection and were not validated, so the reported associations may be unstable. The analysis found associations, not causes; pain self-efficacy and pain severity may be markers of risk rather than targets that, if changed, improve outcomes.

The paper reports funding from the Chartered Society of Physiotherapy Charitable Trust, GP and Physiotherapy Research Facilitators, and the Australian National Health and Medical Research Council.

Do not conclude that older patients or those with longer pain benefit more from being offered direct access to physical therapy. Those subgroup patterns were not significant and had large uncertainty; the analysis found no treatment effect modifier.