Ultrasound-guided tendon-surface injection speeds recovery
Ultrasound-guided tendon-surface injections plus joint injection and home exercise improved early pain and function more than comparison approaches for frozen shoulder.
*Multicenter randomized controlled trial; Level 1 (OCEBM).

Citation

Zhu D, Ou Y, Li Y, et al. A Novel Ultrasound-Guided Injection Strategy for Rapid Functional Recovery in Frozen Shoulder: A Multicenter Randomized Trial. The American Journal of Sports Medicine. 2026;54(6):1324–1332. doi:10.1177/03635465261427347.

Background

Standard non-surgical care for frozen shoulder often helps by about 3 months but may not provide fast relief within 4 weeks. Faster manipulative approaches can work but may cause more harm.

Patients

Adults 40–75 years with one-sided frozen shoulder for 1–12 months, limited lifting to <90° and moderate-to-severe pain. Excluded: full rotator cuff tear, prior shoulder surgery, fracture/dislocation, infection, pregnancy, recent steroid use, bleeding disorder, medication allergy.

Intervention

Ultrasound-guided tendon-surface steroid injections (up to 4 sites) + ultrasound-guided joint injection + home wall-climbing exercise.

Control

(1) Manipulation under local anesthesia + joint injection + exercise; (2) Blind tender-point injections + joint injection + exercise.

Outcome

Primary: 4-week shoulder function on Beijing Friendship Hospital Scale (0–100; higher is better). Secondary: pain scale (0–10), shoulder motion, ultrasound-based shoulder injury score (0–12; higher is worse), complete remission by phone at 1–24 months, harms.

Follow-up Period

1 week and 4 weeks in clinic; phone follow-up to 24 months.

Results

Outcome (time) Tendon-surface vs manipulation Tendon-surface vs blind tender-point
Function score (1 week) +6.0 points (95% CI 3.0 to 9.0) +8.1 points (95% CI 5.1 to 11.0)
Function score (4 weeks) (primary) +7.6 points (95% CI 4.4 to 10.9) +7.9 points (95% CI 4.7 to 11.1)
Pain score (4 weeks) −0.77 (95% CI −1.17 to −0.38) −0.96 (95% CI −1.35 to −0.58)
Complete remission (1 month) +24.3%; NNT 5 +27.3%; NNT 4
Severe pain needing rescue medicine (24 hours) −58.1%; NNT 2 —
Rotator cuff swelling/tear on ultrasound (4 weeks) −55.8%; NNT 2 −12.8%; NNT 8
NNT = number needed to treat for one additional benefit (or one fewer harm).

Limitations

Not blinded, and the trial stopped early after an interim analysis, which can overestimate benefits. The main function scale may not translate directly to other cultures, and the most important “clinically meaningful” point change was not established. Long-term remission was self-reported by phone, without standardized in-person exams or imaging.

Funding

Regional government and military-affiliated hospital innovation grants; no industry sponsor.

Clinical Application

For frozen shoulder needing rapid relief, consider ultrasound-guided tendon-surface injections plus standard joint injection and home exercise; avoid manipulation when safety is a priority.