Comparison of treatments for frozen shoulder: a systematic review and meta-analysis

Key points

1. 01Intra-articular corticosteroid was the only treatment that beat both no treatment or placebo and physiotherapy with results that were clinically and not just statistically meaningful in the short-term.
2. 02Its advantage was clearest for pain and function in the first weeks to 2 to 3 months and appeared to fade by 6 months, when most treatments looked similar.
3. 03Adding arthrographic distension to the injection ranked best for short-term pain in the network analysis but the extra benefit over injection alone was clinically small.
4. 04A simple home exercise program of range-of-motion exercises and stretches gave a clinically meaningful short-term pain benefit on its own.
5. 05The authors recommend offering IA corticosteroid plus a home exercise program at first contact for frozen shoulder of less than 1 year duration, a step earlier than current NICE guidance.

How it was conducted

Design

Systematic review with pairwise and network meta-analysis of randomized trials, following PRISMA; random-effects models and trial sequential analysis (TSA); GRADE for certainty

Search

Medline, EMBASE, Scopus, and CINAHL searched February 2020; randomized designs comparing treatments for frozen shoulder with other treatments, placebo, or no treatment

Intervention

All nonsurgical modalities including IA and subacromial corticosteroid, physiotherapy, home exercise, arthrographic distension, acupuncture, ESWT, oral corticosteroid

Outcomes

Primary outcomes pain (mean difference on VAS) and function (standardized mean difference); secondary outcome external rotation range of movement; follow-up split into short, mid, and long term

Analysis

65 studies in qualitative synthesis, 34 in pairwise meta-analyses, 39 in network meta-analyses; MCRD set at 1 VAS point for pain, effect size 0.45 for function, 10 degrees for ER ROM

What they found

• 65 eligible studies with 4097 participants; 34 studies (2402 participants) in pairwise and 39 studies (2736 participants) in network meta-analyses.
• IA corticosteroid vs no treatment or placebo, short-term pain: MD -1.0 VAS point (95% CI -1.5 to -0.5; P < .001) and function SMD 0.6 (95% CI 0.3 to 0.9; P < .001), both clinically and statistically significant.
• IA corticosteroid vs physiotherapy, late short-term pain: MD -1.1 VAS points (95% CI -1.7 to -0.5; P < .001).
• Mid-term, IA corticosteroid plus home exercise vs no treatment or placebo pain: MD -1.4 VAS points (95% CI -1.8 to -1.1; P < .001); home exercise alone vs no treatment was also clinically and statistically significant.
• TSA confirmed the benefit of IA corticosteroid vs placebo for early and late short-term pain (ruling out a type I error) but did not confirm it vs physiotherapy, suggesting more trials are needed there.
• In the network analysis, arthrographic distension plus IA corticosteroid had the highest probability of being best for late short-term pain (96%), with IA corticosteroid second (85%).

Limitations

• Frozen shoulder of all stages and chronicity was analyzed together, so the best treatment for a specific stage could not be determined.
• Home exercise programs varied in frequency, intensity, and duration but were not adjusted for in comparisons.
• Diverse physiotherapy modalities were grouped and analyzed together despite likely differing effectiveness.
• Certainty of evidence was moderate at best for the key corticosteroid comparisons, and long-term (beyond 12 months) data were insufficient to analyze.

Why it matters

For patients

If you have frozen shoulder that started within the past year, a corticosteroid injection plus simple home stretches may ease pain and improve movement faster than other options in the first few months.

For clinicians

Consider offering IA corticosteroid with a home exercise program at first contact rather than reserving it for treatment failures, while counseling that the advantage is mainly short-term and largely evens out by 6 months.

For readers

This is the largest meta-analysis in the field and supports early corticosteroid injection for frozen shoulder, though benefits are short-lived and the evidence is moderate, not definitive.