This study was a randomized controlled trial of 50 patients, who underwent history taking, physical examinations, x-rays, and MRI or ultrasonography, and were diagnosed with primary adhesive capsulitis of the shoulder from December 2012 to November 2013. The 50 patients were assigned to Group I (anterior approach) or Group II (posterior approach) under double-blinded randomization. The randomization was performed by an independent nurse using a computerized random sequence generator. The operators who participated in the procedure were blinded to the participant's information. The injection mixture for both groups was: 1 mL triamcinolone, 4 mL lidocaine, 4 mL normal saline, and 3 mL Omnipaque (Iohexol, General Electronics Healthcare). After the injection, anteroposterior, scapular lateral, and axillary x-rays of the shoulder were taken of all patients within 15 min, and leakage was classified into three categories by a musculoskeletal radiologist with 18 years experience: 1) no leakage; 2) leakage; and 3) cannot be determined. Injection accuracy was defined as \[number of 1) ÷ {number of 1) + number of 2) + number of 3)}\] × 100 (%). The frequency of needling until the injection had been completed was checked. A visual analog scale (VAS) for pain, range of motion (ROM), patient's satisfaction (SAT), the American Shoulder and Elbow Surgeons (ASES) shoulder score, and the Constant score were recorded before, and 3, 7, and 13 weeks after the injection. VAS scores range from 0 to 10, with 0 being no pain and 10 being the most severe pain ever experienced. Internal rotation was measured from buttock to T5 in this study, with the range between L5 and T5 being allocated a sequential value between 2 and 13 and the range below the first sacral vertebra being allocated 1. SAT was also selected from a scale of 0 to 10, with 0 being unsatisfactory and 10 being very satisfactory. Stick exercises were started at 3 weeks, and theraband exercises were started 7 weeks after the injection.
Two patients of 54 patients who had been assessed for eligibility in this study were excluded due to proximal humerus fractures of the affected shoulders. And another two participants were excluded because they had declined to participate. One Group I patient was excluded due to an additional subacromial injection 3 weeks after the study injection. And another Group I patient was excluded due to loss of follow-up before 13 weeks. Also, two patients were excluded from Group II because of loss to follow-up. Finally, 46 patients were included in the study (n = 23/group). The demographic data of the two groups were not significantly different.
Diagnostic ultrasonography was performed by a musculoskeletal radiologist with 18 years experience using ultrasound (Philips Healthcare) before the injection. The ultrasound-guided glenohumeral steroid injection was administered by a shoulder and elbow surgeon who performed more than 2,000 similar procedures.
A 21-gauge spinal needle was used for all of the ultrasound-guided steroid injections. The patient was supine with the shoulder slightly rotated externally for the anterior approach. The shoulder was draped in sterile fashion using 70% alcohol and povidone-iodine solution. The linear ultrasonography probe (General Electronics Healthcare) was prepared with 70% alcohol and a thin layer of sterile ultrasonographic transmission gel (Republic of Korea). The needle was inserted at the level of the coracoids, from lateral to medial, and was aimed at the medial border of the humeral head, parallel to the longitudinal axis of the probe. The posterior approach was performed with the patient in the lateral decubitus position on the unaffected side, leaning 45 degree anteriorly and hugging a pillow. The needle was inserted from lateral to medial, parallel to the longitudinal axis of the probe, and into the joint between the humeral head and the posterior glenoid labrum.
The participants took off their coats and shirts in the injection room. Females wore a skirt at the axillary level exposing the shoulder for injection. The provider entered the injection room and positioned the patients appropriately. Just after positioning, the operator put on gloves, prepared the injection mixture, and administered the injection. Positioning time (from operator entering the room until the end of positioning) and injection time (from putting on gloves until just after the injection) were recorded by an independent nurse. The sum of the two times was recorded as the total time, and the frequency of needling during the injection for one patient was also recorded.
A power analysis indicated that a sample size of 46 patients (23/group) would provide a statistical power of 80%, with a two-sided α level of 0.05 to detect a significant difference in the improvement of ASES score between initial and 13 weeks after injection, assuming an effect size of 0.85 (mean difference, 16.1; standard deviation, 19.0). This was based on the mean and standard deviation of the improvement of ASES score between initial and 13 weeks after injection observed in a pilot study of 20 patients.
The Wilcoxon signed-rank test was used to compare preoperative and postoperative data. Normally distributed data between the groups were analyzed using the independent sample t-test. Otherwise, the nonparametric Mann-Whitney U-test was used. Statistical analyses were performed with SPSS ver. 13.0 software (SPSS Inc., Chicago, USA). A p value \< 0.05 was considered statistically significant.
