Methods
After obtaining approval from the Institutional Review Board (IRB/EC of Hallym University Chuncheon Sacred Heart Hospital, Chuncheon, Korea; IRB No. 2013-94) and in alignment with the Declaration of Helsinki, a single-center, patient/assessor-blinded, prospective, superiority, randomized controlled trial was conducted from December 2013 to April 2015 with patients aged 35 to 80 who were diagnosed with adhesive capsulitis of the shoulder (ClinicalTrials.gov ID: NCT06493656 , retrospectively registered on July 7, 2024, to allow global access to the study protocol). Patients were consecutively recruited from the outpatient orthopaedic clinic. Those clinically evaluated and diagnosed with adhesive capsulitis were invited to participate and provided informed consent. No advertisements or recruitment activities were conducted outside the clinic. Diagnosis required meeting two or more of the following criteria within the shoulder's range of motion both active and passive: abduction and forward elevation below 100˚, external rotation at the side below 30˚, and internal rotation at 90˚ of shoulder abduction below 30˚ [ 12 , 13 ]. Exclusion criteria included a history of shoulder surgery, trauma, nerve injury, concomitant rotator cuff/biceps lesions, glenohumeral arthritis, calcific tendinosis, cancer, mental illness, hormonal imbalance, or diabetes with HbA1c levels exceeding 7.0%. All patients underwent routine shoulder radiographs, followed by either ultrasound or magnetic resonance imaging, depending on insurance coverage. An independent musculoskeletal fellowship-trained radiologist with 20 years of experience interpreted all magnetic resonance imaging studies (Ingenia 3 T or Achieva 3 T, Philips Healthcare, Bothell, WA) and ultrasound examinations (iU22; 5–12 MHz linear probe, Philips Healthcare, Bothell, WA) to determine patient inclusion and exclusion.
Patients who provided informed consent and remained eligible were randomly assigned using sealed envelopes containing a computer-generated random sequence created with SPSS version 25 (IBM Corporation, Armonk, NY) to receive either corticosteroid injection with hydrodilatation (group CSHD) or corticosteroid injection only (group CS). An independent orthopaedic nurse prepared the sealed envelopes and computer-generated block randomization sequence (block size 4). Patients were blinded to their allocation.
All patients received injections administered by a single shoulder fellowship-trained orthopaedic surgeon within a university hospital setting. Due to differences in injection volumes, the practitioner was not blinded to the study arm. Patients were seated during the procedure, and a 21-gauge spinal needle was inserted into the glenohumeral joint using a posterior approach under ultrasound guidance (Logiq P6 Pro; 5–12 MHz linear probe, GE Healthcare, Chicago, IL).
In the CS group, a total volume of 5 mL was injected intraarticularly, consisting of 1 mL of triamcinolone (Triam inj.; 40 mg/mL triamcinolone acetate, Shinpoong Pharmaceuticals, Seoul, Korea) and 4 mL of 1% lidocaine (Huons Lidocaine HCL inj 1%; 10.67 mg/mL lidocaine hydrochloride hydrate, Huons Pharmaceuticals, Gyeonggi-do, Korea). In the CSHD group, a total injection volume of 20 mL was administered, with 15 mL of normal saline (JW 0.9% Normal Saline inj; sodium chloride 9 mg/mL, JW Pharmaceuticals, Gyeonggi-do, Korea) added to the same 5 mL corticosteroid-lidocaine solution. Capsule preservation was intended by tactile feedback of constant pressure during injection, without a sudden loss of resistance. Sustained dilatation of the joint capsule was further confirmed by the increase in distance between the capsule and the humeral head (Fig. 1 ). Care was taken to perform the injection in the CS group as slowly and consistently as the CSHD group so that patients would not be aware of which treatment option they were allocated to. After injection, a representative ultrasound image demonstrating maintained capsular distension was obtained. Two authors who did not perform the injections independently measured the capsule-humeral head distance on this image. This distance was defined as the longest perpendicular measurement from a tangential line along the humeral head cortex to the inner margin of the joint capsule. The two measurements were averaged for analysis.
Fig. 1 Ultrasound-guided Glenohumeral Hydrodilatation. The needle was introduced into the glenohumeral joint through a posterior approach. The joint capsule ("C"), needle (dotted oval), and humeral head ("H") are visualized A . A 20-mL solution consisting of 1 mL triamcinolone, 4 mL lidocaine, and 15 mL normal saline was injected to perform hydrodilatation. Care was taken to confirm that capsular distension was maintained after injection without sonographic evidence of rupture in the visualized field B . The capsule-humeral head distance was measured as the longest perpendicular distance from a tangential line along the humeral head cortex to the inner margin of the joint capsule (double-sided arrow). The deltoid ("D"), infraspinatus ("I"), and labrum ("L") are also marked in the wider field-of-view image C
Ultrasound-guided Glenohumeral Hydrodilatation. The needle was introduced into the glenohumeral joint through a posterior approach. The joint capsule ("C"), needle (dotted oval), and humeral head ("H") are visualized A . A 20-mL solution consisting of 1 mL triamcinolone, 4 mL lidocaine, and 15 mL normal saline was injected to perform hydrodilatation. Care was taken to confirm that capsular distension was maintained after injection without sonographic evidence of rupture in the visualized field B . The capsule-humeral head distance was measured as the longest perpendicular distance from a tangential line along the humeral head cortex to the inner margin of the joint capsule (double-sided arrow). The deltoid ("D"), infraspinatus ("I"), and labrum ("L") are also marked in the wider field-of-view image C
Patients underwent follow-up assessments at 3 weeks, 7 weeks, 3 months, and 6 months post-injection by an orthopaedic nurse blinded to the study arm. A strict home exercise program was prescribed, consisting of passive stretching exercises for at least 30 min per day (five minutes after waking, after each meal, and ten minutes before sleep). The patient was encouraged to maintain the maximum range of motion in forward elevation, cross-body adduction, and internal rotation, within the maximum tolerable range of pain, using a wand or by leaning against a wall. To support regular stretching exercises, aceclofenac 100 mg (Acecle tab.; SCD Pharmaceuticals, Seoul, Korea) twice daily and omeprazole 20 mg (Opra cap.; Withus Pharmaceuticals, Seoul, Korea) once daily were prescribed as adjunctive analgesic therapy during the initial 3 months.
Patient data, including age, sex, symptom duration, affected side, and hand dominance, were documented. Activity level, based on patients’ profession and participating sports, was recorded on a scale of 1 to 3, where 3 was the most active [ 14 ]. Just before injection and during follow-up, range of motion (ROM) was measured with a goniometer in forward elevation (FE), external rotation at side (ER), and internal rotation at 90˚ of abduction (IR). “Improvement” in ROM was calculated by subtracting the pre-injection measurements from those obtained at each follow-up visit. Clinical outcome was also evaluated using the pain visual analogue scale (pVAS), the American Shoulder and Elbow Surgeons (ASES) score, and the Constant-Murley (Constant) score. Achievement of the minimal clinically important difference for each outcome measure specific to adhesive capsulitis (pVAS, 1.1; ASES, 8.2; Constant, 10.1) was also assessed [ 15 ]. Subjective patient satisfaction regarding the current shoulder status was recorded on a scale from 0 to 100, with 100 indicating the highest level of satisfaction. ROM measurements and collection of clinical scores were performed by an independent nurse who was blinded to the study arm.
The primary outcome, prespecified during the study design based on a pilot study, was improvement in the ASES score at 7 weeks after injection. Other outcomes were analyzed as secondary outcomes. The ASES score was chosen as the primary outcome because it includes domains for pain and activities of daily living requiring functional shoulder motion, thereby addressing both hypotheses regarding ROM and pain improvement. Based on the pilot study of 20 patients, a total of 44 patients were required for this superiority trial to achieve 80% statistical power with a two-sided α level of 0.05 to detect a 15.4 point difference in ASES score improvement from pre-injection to 7 weeks post-injection, assuming a standard deviation of 17.8 between the two groups. Anticipating a 30% dropout rate over the six-month outpatient study period, 31 patients were allocated to each group.
Continuous data normality was assessed using the Kolmogorov–Smirnov test. Between-group comparisons for continuous data were analyzed using either the independent-samples t-test or the Mann–Whitney U test. Intra-group comparisons were conducted using the paired-sample t-test. Results were reported as mean ± standard deviation (SD). Categorical data were analyzed using Pearson's chi-squared or Fisher’s exact tests and presented as percentages. Multiple outcomes assessed at different time points were analyzed using linear mixed-effects models that included treatment group, time, and the group-by-time interaction as fixed effects, with patient-specific random intercepts to account for within-patient correlation. The results are reported with 95% confidence intervals (CIs). No adjustment for multiplicity was performed for repeated secondary outcome analyses across multiple time points. Therefore, these secondary analyses were considered exploratory. Interobserver reliability for capsule-humeral head distance measurements was assessed using the intraclass correlation coefficient (ICC). As the average of the two observers’ measurements was used for analysis, the average-measures ICC was reported. All statistical analyses were performed using SPSS version 25 (IBM Corporation, Armonk, NY). A P-value below 0.05 was considered statistically significant.
Results
A total of 67 patients provided informed consent and were initially assessed for eligibility. Five patients were excluded prior to randomization: four withdrew consent, and one received a corticosteroid injection at another clinic before treatment allocation. The remaining 62 patients were randomized equally into two treatment groups. Eight patients were lost during follow-up, three from the CSHD group and five from the CS group. Additionally, three patients, one in the CSHD group and two in the CS group, developed severe focal acromioclavicular joint pain that required an additional corticosteroid injection into the acromioclavicular joint. One patient in the CSHD group was diagnosed with concomitant subcoracoid impingement syndrome. Consequently, 50 patients, comprising 26 in the CSHD group and 24 in the CS group, were included in the final per-protocol analysis (Fig. 2 ). Despite exclusions, the final per-protocol sample size of 50 patients surpassed the minimum required sample size of 44 patients to achieve 80% power for the primary outcome.
Fig. 2 Flow Diagram of Patient Enrollment
Flow Diagram of Patient Enrollment
Baseline parameters, including demographic data, pre-injection ROM, and clinical scores, did not show significant differences between the groups. However, the post-injection capsule-humeral head distance, ROM in all directions, patient satisfaction, ASES, and Constant scores assessed at the final follow-up after 6 months were significantly better in the CSHD group (Table 1 ). In addition, a significantly greater proportion of patients in the CSHD group exceeded the minimal clinically important difference for the Constant score. Nevertheless, within each group, intra-group paired comparisons between pre-injection and 6 months post-injection demonstrated significant improvements in all parameters (Table 2 ). Interobserver reliability for capsule-humeral head distance measurements was excellent (ICC > 0.90), with an average-measures ICC of 0.955 (95% CI, 0.920–0.974).
Table 1 Demographic, pre-injection, and final follow-up data: inter-group analysis Group CSHD (n=26) Group CS (n=24) P Value
Demographic
Age, mean ± SD, years 54.2 ± 5.2 54.6 ± 8.8 0.820 Symptom duration, mean ± SD, weeks 19.6 ± 14.1 23.8 ± 24.0 0.458 Sex, male: female, n 10: 16 15: 9 0.089 History of diabetes, n (%) 2 (7.7%) 1 (4.2%) 1.000 Activity level, 1: 2: 3, n 9: 8: 9 13: 6: 5 0.354 Laterality, right: left, n 11: 15 6: 18 0.197 Dominant hand affected, n (%) 12 (46.2%) 6 (25.0%) 0.119 Diagnosis, MRI: ultrasound, n 19: 7 15: 9 0.423
Pre-injection
FE ROM, degrees, mean ± SD, ˚ 95.4 ± 13.9 99.6 ± 15.7 0.331 ER ROM, degrees, mean ± SD, ˚ 25.8 ± 8.4 30.0 ± 11.2 0.143 IR ROM, degrees, mean ± SD, ˚ 32.3 ± 4.0 33.1 ± 4.5 0.509 Patient satisfaction, mean ± SD 27.3 ± 14.0 27.1 ± 16.2 0.959 pVAS score, mean ± SD 7.1 ± 1.6 7.0 ± 2.0 0.883 ASES score, mean ± SD 33.3 ± 12.9 36.9 ± 11.3 0.319 Constant score, mean ± SD 42.1 ± 5.7 45.8 ± 8.0 0.071
Post-injection
Capsule-humeral head distance, mm, mean ± SD 7.5 ± 1.2 2.9 ± 1.2 < 0.001*
Final Follow-up
FE ROM, degrees, mean ± SD, ˚ 148.1 ± 4.8 138.8 ± 13.2
0.002*
ER ROM, degrees, mean ± SD, ˚ 74.6 ± 6.9 64.8 ± 9.8 < 0.001* IR ROM, degrees, mean ± SD, ˚ 71.5 ± 9.8 58.1 ± 8.1 < 0.001* Patient satisfaction, mean ± SD 92.3 ± 12.8 79.4 ± 17.0
0.004*
pVAS score, mean ± SD 0.7 ± 1.1 1.2 ± 1.2 0.138 Patients exceeding MCID, n (%) 26 (100.0%) 23 (95.8%) 0.480 ASES score, mean ± SD 88.5 ± 11.6 81.8 ± 10.4
0.040*
Patients exceeding MCID, n (%) 26 (100.0%) 24 (100.0%) N/A** Constant score, mean ± SD 63.8 ± 7.4 57.7 ± 8.6
0.011*
Patients exceeding MCID, n (%) 24 (92.3%) 12 (50.0%) < 0.001* CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; SD, standard deviation; MRI, magnetic resonance imaging; FE, forward elevation; ROM, range of motion; ER, external rotation at side; IR, internal rotation at 90˚ of shoulder abduction; pVAS, pain visual analogue scale; ASES, American Shoulder and Elbow Surgeons; Constant, Constant-Murley. * P value < 0.05. ** N/A, not applicable as all patients in both groups exceeded the MCID for the ASES score
Demographic, pre-injection, and final follow-up data: inter-group analysis
CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; SD, standard deviation; MRI, magnetic resonance imaging; FE, forward elevation; ROM, range of motion; ER, external rotation at side; IR, internal rotation at 90˚ of shoulder abduction; pVAS, pain visual analogue scale; ASES, American Shoulder and Elbow Surgeons; Constant, Constant-Murley. * P value < 0.05. ** N/A, not applicable as all patients in both groups exceeded the MCID for the ASES score
Table 2 Pre-injection versus final follow-up in 6 months: intra-group analysis Pre-injection Final follow-up P Value Group CSHD (n = 26) FE ROM, degrees, mean ± SD, ˚ 95.4 ± 13.9 148.1 ± 4.8 < 0.001* ER ROM, degrees, mean ± SD, ˚ 25.8 ± 8.4 74.6 ± 6.9 < 0.001* IR ROM, degrees, mean ± SD, ˚ 32.3 ± 4.0 71.5 ± 9.8 < 0.001* Patient satisfaction, mean ± SD 27.3 ± 14.0 92.3 ± 12.8 < 0.001* pVAS score, mean ± SD 7.1 ± 1.6 0.7 ± 1.1 < 0.001* ASES score, mean ± SD 33.3 ± 12.9 88.5 ± 11.6 < 0.001* Constant score, mean ± SD 42.1 ± 5.7 63.8 ± 7.4 < 0.001* Group CS (n = 24) FE ROM, degrees, mean ± SD, ˚ 99.6 ± 15.7 138.8 ± 13.2 < 0.001* ER ROM, degrees, mean ± SD, ˚ 30.0 ± 11.2 64.8 ± 9.8 < 0.001* IR ROM, degrees, mean ± SD, ˚ 33.1 ± 4.5 58.1 ± 8.1 < 0.001* Patient satisfaction, mean ± SD 27.1 ± 16.2 79.4 ± 17.0 < 0.001* pVAS score, mean ± SD 7.0 ± 2.0 1.2 ± 1.2 < 0.001* ASES score, mean ± SD 36.9 ± 11.3 81.8 ± 10.4 < 0.001* Constant score, mean ± SD 45.8 ± 8.0 57.7 ± 8.6 < 0.001* CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; SD, standard deviation; FE, forward elevation; ROM, range of motion; ER, external rotation at side; IR, internal rotation at 90˚ of shoulder abduction; pVAS, pain visual analogue scale; MCID, minimal clinically important difference; ASES, American Shoulder and Elbow Surgeons; Constant, Constant-Murley. * P value < 0.05
Pre-injection versus final follow-up in 6 months: intra-group analysis
CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; SD, standard deviation; FE, forward elevation; ROM, range of motion; ER, external rotation at side; IR, internal rotation at 90˚ of shoulder abduction; pVAS, pain visual analogue scale; MCID, minimal clinically important difference; ASES, American Shoulder and Elbow Surgeons; Constant, Constant-Murley. * P value < 0.05
Improvement in the ASES score at week 7, the primary outcome, was significantly greater in the CSHD group by 10.51 points (95% CI, 4.12–16.89; p = 0.001) (Table 3 ). At 3 weeks post-injection, improvements in pVAS ( p = 0.009) and ASES score ( p < 0.001) were significantly greater in the CSHD group. At week 7, improvements in FE ROM ( p = 0.045), ER ROM ( p = 0.041), IR ROM ( p = 0.017), pVAS ( p = 0.027), ASES score ( p = 0.001), and Constant score ( p < 0.001) were significantly greater in the CSHD group. At 3 months, all outcomes except pVAS showed significantly greater improvement in the CSHD group, whereas improvement in pVAS was not significantly different between the groups ( p = 0.066). Similar findings were observed in 6 months, with no significant between-group difference in pVAS improvement ( p = 0.136) but significantly greater improvement in all other outcomes in the CSHD group (Fig. 3 ).
Table 3 Comparison of the amount of improvement from pre-injection: inter-group analysis β (95% CI), P value CSHD vs. CS Δ FE ROM Δ ER ROM Δ IR ROM Δ Patient Satisfaction Δ pVAS Δ ASES Δ Constant 3wks–Pre 3.30 (− 3.42- 10.02), 0.334 2.95 (− 2.40–8.30), 0.278 1.65 (− 2.30–5.60), 0.411 4.92 (− 3.49–13.33), 0.250 − 1.04 (− 1.82–0.27), 0.009* 11.24 (4.86–17.63), <0.001* 2.82 (− 0.28–5.92), 0.074 7wks–Pre 6.88 (0.16–13.59), 0.045* 5.58 (0.23–10.93), 0.041* 4.82 (0.87–8.77), 0.017* 6.44 (− 1.96–14.85), 0.132 − 0.88 (− 1.65–0.10), 0.027* 10.51 (4.12–16.89), 0.001* 5.28 (2.18–8.38), <0.001* 3mos–Pre 10.74 (4.02–17.46), 0.002* 12.85 (7.50–18.20), <0.001* 9.82 (5.87–13.77), <0.001* 12.26 (3.85–20.67), 0.004* − 0.73 (− 1.50–0.05), 0.066 8.95 (2.57–15.34), 0.006* 7.72 (4.62–10.82), <0.001* 6mos–Pre 13.53 (6.81–20.25), <0.001* 14.05 (8.71–19.40), <0.001* 14.23 (10.28–18.18), <0.001* 12.71 (4.30–21.11), 0.003* − 0.59 (− 1.37–0.19), 0.136 10.26 (3.88–16.65), 0.002* 9.81 (6.71–12.91), <0.001* Δ Amount of Improvement = Follow-up measurement subtracted by the pre-injection measurement β = Amount of improvement in the CSHD group subtracted by that of the CS group CI, confidence interval; CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; SD, standard deviation; FE, forward elevation; ROM, range of motion; ER, external rotation at side; IR, internal rotation at 90˚ of shoulder abduction; pVAS, pain visual analogue scale; ASES, American Shoulder and Elbow Surgeons; Constant, Constant-Murley. * P value < 0.05. In all cases, the values were favorable in the CSHD group
Comparison of the amount of improvement from pre-injection: inter-group analysis
Δ Amount of Improvement = Follow-up measurement subtracted by the pre-injection measurement
β = Amount of improvement in the CSHD group subtracted by that of the CS group
CI, confidence interval; CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; SD, standard deviation; FE, forward elevation; ROM, range of motion; ER, external rotation at side; IR, internal rotation at 90˚ of shoulder abduction; pVAS, pain visual analogue scale; ASES, American Shoulder and Elbow Surgeons; Constant, Constant-Murley. * P value < 0.05. In all cases, the values were favorable in the CSHD group
Fig. 3 Serial Measurements from Pre-injection to Last Follow-up. A Forward Elevation. B External Rotation at Side. C Internal Rotation at 90˚ of Shoulder Abduction. D Pain Visual Analogue Scale. E American Shoulder and Elbow Surgeons (ASES) Score. F Constant-Murley Score. CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only. * P value < 0.05
Serial Measurements from Pre-injection to Last Follow-up. A Forward Elevation. B External Rotation at Side. C Internal Rotation at 90˚ of Shoulder Abduction. D Pain Visual Analogue Scale. E American Shoulder and Elbow Surgeons (ASES) Score. F Constant-Murley Score. CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only. * P value < 0.05
No procedure-related adverse events were observed during the study. No patients experienced side effects related to injected corticosteroids or oral medications. The additional acromioclavicular joint injections and the diagnosis of concomitant subcoracoid impingement syndrome were considered post-randomization clinical events or alternative pain generators rather than adverse events directly attributable to the assigned intervention. No treatment-related discontinuations occurred.
Background
Adhesive capsulitis of the shoulder is a common condition mostly treated conservatively, but it can severely decrease the quality of life in those affected. The disease primarily presents as a limited active and passive range of motion, with associated discomfort at the end range. Sleep disturbance and weakness due to pain are also common [ 1 , 2 ].
Glenohumeral corticosteroid injection has been one of the well-established treatments for adhesive capsulitis. It is especially beneficial in the initial “freezing” stage of the disease as it works against overt synovial inflammation and the resultant capsular thickening, therefore decreasing the duration and severity of the disease [ 3 ]. Hydrodilatation, which aims to stretch or rupture the joint capsule using hydraulic pressure, is also a viable treatment option with results comparable to steroid injection when done in isolation [ 4 ].
The combined treatment of the above two options can easily be achieved by adding corticosteroids to the hydrodilatation solution [ 5 ]. However, only a number of randomized controlled trials have directly compared this combined treatment versus isolated corticosteroid injection, and the study protocol and conclusions are heterogeneous [ 6 – 11 ]. Some studies involve fluoroscopic visualization of capsular rupture or, on the contrary, ultrasound confirmation of the capsule being intact. The amount of solutions injected for hydrodilatation varies as well. Furthermore, as these are outpatient-based studies, the follow-up period is relatively short, mostly under three to four months.
Therefore, this study aims to validate a streamlined, simple protocol of hydrodilatation combined with corticosteroid injection by comparing it with isolated glenohumeral corticosteroid injection through a six-month randomized controlled trial.
The primary hypothesis is that due to the dilatation effect, the recovery of ROM after combined hydrodilatation with corticosteroid injection will be faster and more evident than that of isolated corticosteroid injection. The secondary hypothesis is that early pain reduction may be prominent in isolated corticosteroid injections, as concomitant hydrodilatation may dilute the injected corticosteroid.
Conclusion
In treating adhesive capsulitis, adding saline to a conventional corticosteroid injection to perform a single-shot, capsule-preserving hydrodilatation with a 20 mL solution may improve short- to mid-term functional recovery compared to an isolated corticosteroid injection.
Discussion
The key finding of this study is that single-shot capsule-preserving hydrodilatation combined with corticosteroid injection resulted in earlier clinical improvement compared with corticosteroid injection alone. This combined approach also demonstrated statistically significantly better range of motion, ASES scores, and Constant scores over a six-month follow-up period. Among these statistically significant differences, MCID analysis suggested a clinical advantage for the Constant score. Pain relief occurred significantly earlier after combined hydrodilatation, with greater improvement in pain VAS than after isolated corticosteroid injection up to 7 weeks. However, at the final 6-month follow-up, pain VAS scores were comparable between the two groups. Therefore, the primary hypothesis was accepted, while the secondary hypothesis was not.
Multiple randomized controlled trials have similarly compared isolated glenohumeral corticosteroid injection to concomitant hydrodilatation with corticosteroid injection in treating adhesive capsulitis (Table 4 ) [ 6 – 11 ]. However, the number of injections, injection volume, interval, regimen, injection guidance, and capsular status varied across studies. The follow-up period also differed, with some studies reporting final outcomes as early as 8 weeks. These methodological differences may explain the heterogeneous results, with some studies demonstrating a significant additional benefit of combined hydrodilatation and others showing no clear advantage.
Table 4 Previous level 1 randomized controlled trials comparing CSHD versus CS References Hydrodilatation Volume (mL) Number of injections and interval Guidance Capsule status Follow-up duration Conclusion Jacobs [ 9 ] 10 3, 6 weeks Landmarks Unstated 12 weeks CSHD better Gam [ 11 ] 19 Up to 6, 1 week Ultrasound Unstated 12 weeks CSHD better Yoon [ 10 ] 49 Once Arthrogram Ruptured 6 months CSHD better Tveitå [ 8 ] 20 3, 2 weeks Arthrogram Ruptured 6 weeks Both similar Sharma [ 6 ] 8 to 20 4, 1–2 weeks Landmarks Preserved 8 weeks * Both similar Lee [ 7 ] 25.1 ± 6.1 Once Ultrasound Preserved 12 weeks Both similar CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; * Additional one-year follow-up evaluation done with a telephone interview
Previous level 1 randomized controlled trials comparing CSHD versus CS
CSHD, glenohumeral corticosteroid injection with hydrodilatation; CS, glenohumeral corticosteroid injection only; * Additional one-year follow-up evaluation done with a telephone interview
Jacobs et al. utilized a 10 mL dilatation solution (triamcinolone 40 mg/1 mL, bupivacaine 6 mL, air 3 mL) in three blind injections at six-week intervals [ 9 ], Gam et al. injected a solution of 19 mL (triamcinolone 20 mg, lidocaine 19 mL) under ultrasound guidance once a week for a maximum of 6 weeks until the symptoms subsided [ 11 ]. Yoon et al. performed a one-time injection with a 49 mL solution (triamcinolone 40 mg/1 mL, lidocaine 4 mL, contrast medium 4 mL, normal saline 40 mL) under ultrasound and confirmed capsular rupture with fluoroscopy [ 10 ]. Despite their varied treatment protocols, the above three studies collectively demonstrated that concomitant hydrodilatation was significantly more effective than corticosteroid injection alone.
Nevertheless, the following studies suggest otherwise, demonstrating no advantage in hydrodilatation over conventional corticosteroid injection alone. Tveitå et al. [ 8 ] administered three 20-mL injections (triamcinolone 20 mg/2 mL, contrast medium 4 mL, bupivacaine 4 mL, normal saline 10 mL) in two-week intervals under fluoroscopic guidance, confirming capsular rupture. Sharma et al. conducted four hydrodilatations within one month using a solution ranging from 8 to 20 mL (triamcinolone 20 mg/1 mL, lidocaine 3 mL, and 4 to 16 mL normal saline). The maximum volume endpoint was determined by either reaching the maximum pressure tolerable by the patient or the onset of intolerable pain [ 6 ]. Lastly, Lee et al. performed a slow (0.25 mL/sec) and careful “capsule-preserving” hydrodilatation. The solution comprised triamcinolone 40 mg/1 mL, lidocaine 3 mL, and a variable amount of normal saline with a total volume of 25.1 ± 6.1 mL until no further injection could be injected manually while maintaining capsule distension [ 7 ].
The findings of the present study may not be new, as several high-level meta-analyses concluded that concomitant hydrodilatation with corticosteroids is a better treatment option than isolated corticosteroid injection [ 5 , 16 , 17 ]. Of note, the follow-up period in this study was 6 months, which is beyond most of the above studies, which ranged from 6 to 12 weeks. Most importantly, the novelty of the current study lies in its technical simplicity. Unlike other studies, the technique used in this study did not require fluoroscopic guidance, multiple injections, or confirmation of capsular rupture. It is simple and resource-efficient, as only normal saline was added to the conventional ultrasound-guided one-time corticosteroid injection to create an injection volume of 20 mL. Furthermore, though ultrasound was used in this study to confirm capsular distension, a bony landmark-based anterior injection through the rotator interval may further simplify the process while achieving similar results [ 18 ].
Whether capsular rupture is necessary, beneficial, or possible at all is still undetermined. Capsular thickening, notably the inferior capsule, is a hallmark of adhesive capsulitis, as confirmed by arthroscopic and magnetic resonance imaging [ 19 , 20 ]. The thorough release of the inferior capsule can sometimes be challenging due to joint contracture and necessitates a step-by-step release of the rotator interval, the middle glenohumeral ligament, distally to the inferior glenohumeral ligament, and the axillary pouch. Furthermore, thickened fibrous tissue often requires a combination of punch forceps and radiofrequency ablators for full-thickness resection [ 19 ]. Reports of iatrogenic fracture during manipulation under anaesthesia also indicate that these thickened capsules can be stronger than bone, especially in cases of old age and disuse osteopenia [ 21 , 22 ].
Therefore, it is questionable to claim that the capsule is "ruptured" solely by manual injection, regardless of the amount of dilatant fluid. A more plausible explanation could be that the injection fluid "leaked" through a small puncture at a weak point in the capsule. Such occurrences may not only reduce dilatation pressure but also cause extra-articular dispersion of therapeutic corticosteroids. This issue can be exacerbated when the corticosteroid is diluted with excessive amounts of normal saline, resulting in only a portion of the therapeutic dose remaining within the joint. The varying conclusions of the studies, as mentioned earlier, may be attributed to this phenomenon.
Previous studies on hydrodilatation have explored variable volumes of additional saline to preserve capsular integrity. However, the injection endpoint in these studies depended on either the patient’s perception of intolerable pain or the practitioner’s tactile feedback [ 6 , 7 ]. Moreover, capsular rupture was reported in 6.3% of cases despite efforts to preserve the capsule, even with a very slow infusion rate of 0.25 mL/sec [ 7 ]. The current study aimed to streamline the hydrodilatation process by administering a fixed volume of 20 mL under conventional ultrasound guidance without fluoroscopy. This volume was chosen to achieve glenohumeral joint distension while minimizing the likelihood of capsular rupture. In all cases, the full volume was successfully injected despite resistance, and ultrasound confirmed maintained capsular distension.
Although glenohumeral joint volume varies among individuals, it has been reported to range from 28 to 35 mL in the normal population, as determined by arthrography [ 23 ]. In addition, a 3D volumetric MR arthrography study reported that shoulder joint capacity in patients with adhesive capsulitis, including the rotator interval and bicipital groove, was 22.5 ± 1.1 cm 3 [ 24 ]. These findings support the rationale for using a 20 mL injection volume to achieve capsular distension while reducing the risk of rupture and allowing potential distribution of the solution throughout the glenohumeral joint, including the rotator interval and bicipital groove.
Nevertheless, this study has limitations. First, the number of enrolled patients for per-protocol analysis is relatively small. Although statistically significant results were obtained, the study may have been underpowered to detect differences in final pVAS improvement. In addition, despite randomization, baseline trends were observed between the groups in sex, affected side, dominant side, and pre-injection Constant scores. Therefore, potential confounding from these baseline differences cannot be completely excluded. Second, an intention-to-treat analysis, which is generally considered the preferred analytical strategy for randomized controlled trials, was not performed. Reliance on per-protocol analysis may introduce bias by excluding randomized patients and may limit the robustness and generalizability of the findings. However, given the small number of exclusions and their clear clinical justification, the per-protocol analysis was retained as the primary analysis to avoid dilution of treatment effects from post-randomization clinical events or alternative pain generators. Third, adherence to oral medication and self-directed stretching exercises could not be monitored. As the literature suggests, comparable effectiveness between supervised therapy and self-exercise [ 25 ], future studies could explore the use of telehealth and smartphone applications for better patient compliance [ 26 ]. As the study took place in a rural-suburban community, regular access to formal physiotherapy was limited. Nevertheless, both groups showed significant improvement, with evident advantages in the CSHD group. Fourth, capsular integrity was indirectly assessed through tactile feedback and ultrasound rather than a direct evaluation using contrast media and fluoroscopy. Yet, this approach was not feasible given the study's aim to validate a streamlined hydrodilatation process. Fifth, the surgeon could not be blinded because of the different injection volumes, representing an inherent source of performance bias. To minimize other potential sources of bias, post-injection evaluations and data analyses were performed by independent personnel who were blinded to the treatment allocation. Sixth, the stage of adhesive capsulitis was not stratified among the participants. Although the hallmark features of the three stages can be theoretically distinguished, objective classification was difficult because symptoms often overlap considerably. However, disease duration was recorded, and no significant difference was observed between the groups. Lastly, this study was conducted at a single center and included only Asian patients, which may limit the generalizability of the findings. The retrospective registration of the study protocol in the ClinicalTrials.gov registry should also be acknowledged as a limitation regarding transparency.
Supplementary Material
Below is the link to the electronic supplementary material.
Supplementary Material 1
Supplementary Material 1
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.