Skip to main content

Research Repository

Advanced Search

Primary care decision-making for shoulder pain: identifying treatment effect moderators using clinical expertise

McRobert, Cliona

Primary care decision-making for shoulder pain: identifying treatment effect moderators using clinical expertise Thumbnail


Cliona McRobert


Shoulder pain is a common, costly condition with variable prognosis. Commonly used treatments for shoulder pain in primary care include: (i) advice & analgesia, (ii) exercise and/or manual therapy, and (iii) corticosteroid injection. Current guidelines do not assist clinicians in optimal treatment selection for this condition. Prognostic factors help identify subgroups likely to have poor prognosis, however their potential to help clinicians decide between different treatments is unclear.

A systematic review identified which patient attributes modify effects of these three treatments. Clinical consensus workshops were conducted with 21 UK-based clinicians who manage shoulder pain to identify patient attributes relevant to treatment decision making.The impact of these attributes on treatment choice was studied in a conjoint analysis study of decision-making for shoulder pain.

The review identified 20 potential treatment effect moderators, with low quality evidence.Clinical consensus workshops identified 12 salient patient attributes. The conjoint study received responses from 387 clinicians (31 countries, 64% UK). Results showed that 11 of the 12 attributes discriminated between treatment choices, following adjustment for responders’ country, profession, and experience. Recommending injection was most strongly associated with lack of improvement (OR 2.81, 95%CI 2.16; 3.65), previous positive response to injection (2.79, 2.07; 3.76), and patient preference (2.41, 1.82; 3.19).Recommending physiotherapy was most strongly influenced by patient preference (2.77,2.16; 3.55), presence of weakness/instability (2.05, 0.79; 1.23) and previous positive response to physiotherapy (2.22, 1.76; 2.80). Not recommending corticosteroid injection was associated with traumatic onset and unstable diabetes or cardiac issues, whereas not recommending physiotherapy was associated with sleep disturbance and high pain.

The relative importance of patient attributes that influence shoulder treatment selection was quantified. Logical clinical patterns emerged suggesting that specific patient attributes guide clinicians treatment selection. Future research is indicated to assess if identified attributes indeed modify treatment effects.

Publication Date Jun 1, 2018


Downloadable Citations