Identifying potential moderators of first-line treatment effect in patients with musculoskeletal shoulder pain: a systematic review

Abstract

Background

Commonly used conservative shoulder pain treatments include: advice/analgesia, exercise/manual therapy and, corticosteroid injection. Moderators, patient/clinical attributes influencing treatment effect, facilitate clinical decision-making by identifying which patients might respond best to specific treatments. This review summarises results of studies aiming to identify/test treatment effect moderators.

Methods

Randomised controlled trials (RCTs) containing some form of, or suggested moderation/subgroup analysis (sample size >20, and >10 subjects in smallest subgroup), comparing above treatments against physical/functional/pain outcomes, in adults with shoulder pain were searched for in Medline, Embase, PsychInfo, CINAHL, AMED, Pedro, Cochrane Database. Cochrane Risk of Bias tool and Pincus criteria for moderation analysis were applied.

Results

Six RCTs aiming to identify/test moderators and 16 suggesting potential moderators were included and data narratively synthesised. One trial offered confirmatory level moderation (Pincus criteria). Graded exercise had smaller effect in those with painful arc at baseline, compared against without, although lacked statistical significance (mean difference −14.0 shoulder disability (0–100 scale), 95% CI’s [−28.1, 0.1], p = 0.05). Twenty other factors with insufficient level moderation evidence were identified.

Discussion

Review highlights lack of high-quality evidence for moderators of treatment effect of shoulder pain treatments. Future research should address proposed candidate moderators, using robust moderation methodologies to inform clinical decision-making.

Keywords:

• Shoulder
• physiotherapy
• primary care
• systematic review
• clinical reasoning
• evidence based physiotherapy/medicine
• EBM
• methodology

Acknowledgements

The authors thank Dr Miriam Hattle, statistician, for her careful reading and helpful advice when extracting information regarding the statistical analysis (moderation) methods used in each of the trials included in this review.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was funded by Arthritis Research UK, National Institute for Health Research – School for Primary Care Research, PhD Studentship award to Cliona McRobert, Programme Grants for Applied Research [RP-PG-0615-20002].