A decision-support tool cut poor musculoskeletal care — when GPs used it

Design

Mixed-methods cluster-randomised trial

Population

19 UK general practices; 10,182 adults with musculoskeletal consultations

Primary outcome

Records with at least one of 12 non-evidence-based management options

Effect

3% vs 25% when the tool was used; aOR 12.30 (95% CI 2.52–59.9)

Musculoskeletal problems make up about one in seven UK general practice consultations. This cluster-randomised trial gave 10 of 19 practices three months' access to a web-based decision-support tool for these consultations, after training.

Uptake was the main finding. The tool was opened in 365 of 4,268 eligible patients (8.6%). Where it was used, blinded review found at least one of 12 non-evidence-based management choices in 3% of records, against 25% in control practices (adjusted OR 12.3, 95% CI 2.5–59.9). Tool-guided consultations included more self-management advice and community referral, and less prescribing and fewer fit notes. Across all eligible patients, only self-management advice increased.

The per-protocol comparison is vulnerable to selection: clinicians may have opened the tool for straightforward cases. The very wide interval reflects only 60 records per arm. Safety concerns were logged in 6.6% of uses, none high-risk. The lesson for practice is less about this product than about the pattern it exposes: avoidable imaging, opioids and passive treatments remain common in musculoskeletal care.

• Lead musculoskeletal consultations with self-management advice and activity, which the tool most consistently increased.
• Avoid routine imaging for non-specific back and joint pain without red flags.
• Consider community physiotherapy referral before repeat analgesic prescriptions.
• A decision aid only helps if it fits the consultation; uptake here was under one in ten.

Why it matters

Poor musculoskeletal care was common in control practices — one record in four.

Don't overread it

The large benefit is per-protocol in 8.6% of cases; across all patients only self-management advice changed.

The statistics, in plain English

An odds ratio of 12.3 sounds enormous, but it rests on 2 versus 15 records out of 60 each, so the 95% confidence interval runs from 2.5 to 59.9. The per-protocol comparison counts only consultations where the tool was opened, which may have been easier cases.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.