Ultrasound-guided hydrodilatation combined with corticosteroid injection versus corticosteroid injection alone for frozen shoulder: a retrospective cohort study

To compare 6-month pain, range of motion (ROM), and functional outcomes after ultrasound-guided hydrodilatation combined with corticosteroid injection (HCCI) versus corticosteroid injection alone in patients with primary frozen shoulder, and to explore whether baseline ultrasound parameters are associated with treatment response. This retrospective, non-randomized cohort study used routinely collected clinical records of 215 patients with primary unilateral frozen shoulder treated between January 2021 and December 2025 (database lock, June 30, 2026). Patients were classified according to the treatment received in routine care: HCCI ( n = 124) or ultrasound-guided intra-articular corticosteroid injection alone ( n = 91). The primary outcome was improvement in Shoulder Pain and Disability Index (SPADI) total score at 6 months. Secondary outcomes were changes in ROM, Visual Analog Scale (VAS) pain scores, and SPADI subscales. Adjusted linear regression models were fitted for change scores, with adjustment for the baseline value of the same outcome, age, sex, diabetes mellitus, symptom duration, posterior joint capsule thickness, coracohumeral ligament thickness, and glenohumeral joint space distance. Poor functional outcome (SPADI improvement < 20 points) was compared between groups and examined in an exploratory Firth penalized logistic regression, and inverse probability of treatment weighting (IPTW) was used as a sensitivity analysis. Baseline demographic, clinical, and ultrasound characteristics were well balanced, with all standardized mean differences < 0.10. In the adjusted primary analysis, HCCI was associated with greater SPADI total improvement than corticosteroid injection alone (adjusted mean difference [aMD] = 3.29 points; 95% CI, 1.60 to 4.99; P < 0.001); this between-group difference is smaller than published SPADI minimal clinically important difference estimates. HCCI was also associated with greater ROM improvement, most notably for active abduction (aMD = 13.92°; 95% CI, 9.53 to 18.32) and passive abduction (aMD = 10.53°; 95% CI, 6.18 to 14.88), although no established threshold for clinically meaningful change in ROM was applied. Adjusted VAS differences favored HCCI but were small (aMD range, 0.33 to 0.39 cm). Poor functional outcome was infrequent, occurring in 8/91 patients (8.8%) in the corticosteroid injection group and 3/124 (2.4%) in the HCCI group (risk difference, − 6.4% points; 95% CI, − 14.1 to − 0.2; Fisher’s exact P = 0.057). With only 11 events, the adjusted analysis of this outcome was exploratory and inconclusive (Firth penalized OR = 0.25; 95% CI, 0.05 to 1.21; P = 0.085) and should be regarded as hypothesis-generating only. The IPTW analysis was consistent with the primary analysis (aMD = 3.25; 95% CI, 0.36 to 6.13). Baseline ultrasound thickness parameters were not independently associated with 6-month SPADI. In this retrospective cohort, HCCI was associated with greater 6-month improvement in ROM and with statistically significant but modest additional improvement in pain and SPADI scores compared with corticosteroid injection alone. Because treatment was not randomized, these associations cannot be interpreted as causal effects, and the clinical relevance of the small pain and function differences is uncertain. Prospective randomized studies with standardized treatment allocation, adverse-event monitoring, adherence assessment, and longer follow-up are needed.