Shoulder hydrodilatation for primary, post-traumatic and post-operative adhesive capsulitis

Abstract

Background

Adhesive capsulitis (frozen shoulder) is characterised by pain and loss of range of motion of the glenohumeral joint. It can be present as primary (idiopathic) or secondary to surgery, trauma or other conditions that restrict the use of the shoulder joint. Various treatment options have been adopted including physiotherapy, manipulation under anaesthetic, hydrodilatation and arthroscopic or open capsular release but the optimal form of management remains uncertain.

Objectives

The purpose of the study was to assess the clinical outcome of glenohumeral hydrodilatation in three cohorts of patients with different aetiologies with adhesive capsulitis.

Study design & methods

We carried out a retrospective study of patient who underwent hydrodilatation for adhesive capsulitis between 2013 and 2015. The procedure was performed by a specialist musculoskeletal radiologist under radiological guidance. The injection consisted of steroids, local anaesthetics and NaCl solution with a target volume around 35 mL. Our outcome measures were range of motion, and pre- and post-operative pain. Patients were divided into three groups based on the presumed cause of their stiffness: idiopathic, post-traumatic and post-surgical.

Results

Two hundred fifty patients were included, with a mean age of 59 years (range: 20–79). Of these, 180 had idiopathic primary adhesive capsulitis (27 were diabetic), 23 were post-traumatic, and 20 following surgical procedures. Thirty-four required further intervention following initial hydrodilatation with 8 undergoing repeat hydrodilatation, and 26 requiring arthroscopic capsular release. The diabetic group accounted for 16 capsular releases and 4 repeat procedures, while the idiopathic group accounted for 9 and 4, respectively. One patient required capsular release in the surgical group. An improvement was recorded in ROM in all groups with mean abduction improving from 59° to 110°, flexion from 50° to 120° and external rotation from 20° to 50°. With regards to pain, the majority showed an improvement from severe or moderate pain to no or mild pain.

Conclusions

Results show that hydrodilatation resulted in an improvement in all outcome measures, with only a small number of patients, especially those with diabetes, needing further procedures or showing no improvement in range of motion and pain. There was no difference between the post-traumatic and post-surgical groups.

Introduction

Adhesive capsulitis is one of the most common disorders affecting the glenohumeral joint. It presents as pain with progressive loss of the range of motion (ROM). ¹ Primary or idiopathic adhesive capsulitis is defined by the absence of any intrinsic pathology of the glenohumeral joint and the aetiology of which is not yet well understood. In 1945, Neviaser concluded that the main pathology of this condition is increased thickening and adhesion of the capsule to the humeral head causing pain and restriction of passive and active movement. ² Histologically, fibroblasts with abundant type III collagen are present. ³

Secondary adhesive capsulitis can ensue from surgery, trauma ⁴ or other conditions that restrict the use of the shoulder joint such as stroke, osteoarthritis, prolonged hospitalisation and axillary dissection due to malignancy. ⁵ Diagnosis is clinical but plain radiographs will identify any underlying fractures of osteoarthritis or tendon calcification which can present in a similar manner. Neviaser described the decrease of joint volume with obliteration of axillary recess and subscapular bursa as radiological features of adhesive capsulitis. ⁶

Adhesive capsulitis is considered to be a self-limiting condition comprising three phases: inflammatory/freezing, frozen and thawing with each persisting for a variable period of time.^7,8 The frozen phase can be subdivided into the pain predominant and stiffness predominant phases. Although pain and stiffness might resolve over months to years without intervention ⁹ various management options are considered to help improve the pain and function. Pain medication and physiotherapy is first-line initial treatment, followed by intra articular injection, subacromial injection and injection with hydrodilatation (HD).^10,11 More invasive procedures such as manipulation under anaesthesia (MUA) and arthroscopic capsular release^12,13 have both shown to be effective in cases when non-operative means have failed. There is no consensus within the literature as to which treatment option is optimal.^14,15

Primary adhesive capsulitis is a diagnosis of exclusion; in order to reach such diagnosis thorough examination and investigation need to be done; many overlapping conditions can mimic the presentation of pain and limitation in the ROM of the shoulder. ¹⁶

HD is an arthrographic distension of the glenohumeral joint capsule done under ultrasound or fluoroscopic guidance causing stretch or rupture of the capsule with subsequent increase in ROM. This was introduced as an injection treatment modality by Andrén and Lundberg in 1965. ¹⁷ A randomised controlled trial was done in 2016 to compare the three injection modalities used to treat frozen shoulder (FS) showed the superiority of HD at three months follow-up over the other types of injection. However, six months follow-up showed similar result among them in regards to pain and ROM. ¹⁸

The aim of this study specifically was to assess the clinical outcome of HD carried out on patients with idiopathic, post-traumatic and post-operative adhesive capsulitis.

Patients and methods

The clinical outcomes of 250 patients who underwent HD between 2013 and 2015 at a tertiary orthopaedic centre were reviewed. Patients were divided into three groups based on the aetiology of adhesive capsulitis: idiopathic (primary), post-traumatic (following proximal humerus fractures irrespective of whether received surgical fixation or not) and postoperative (only following elective procedures). Furthermore, we analysed diabetic patients as a sub-group. All patients were referred to HD after failing physiotherapy. The ROM and pain scores for each patient were recorded before the HD and then at three months following the procedure. Final outcome was improvement in ROM at one year. Pain was rated as none, mild, moderate and severe. The ROM related to flexion, abduction and external rotation. Internal rotation was also noted on clinical examination but was not included in the outcome measures as it was difficult to objectively quantify. We assessed the difference in ROM before and after HD based on the aetiology of adhesive capsulitis and the diabetic status of patients.

All of the HD procedures were carried out by a musculoskeletal radiologist using a standardised protocol. The patient is positioned supine and the glenohumeral joint is accessed via the rotator interval anteriorly under fluoroscopic control. A 21-gauge needle is inserted vertically into rotator interval and connected to a set with a three-way tap. A 5 mL of iodinated Omnipaque contrast is then injected to confirm intra-articular position. The arthrogram confirms the low volume of the joint with reduced axillary recess volume and excludes rotator cuff tears. Forty milligrams of Kenalog with 10 mL of 0.25% levobupivacaine are injected via a three-way tap opening and closing to maintain pressure. Finally, 20 mL of 0.9% NaCl solution are injected to distend in an inject and relax technique. The procedure is complete when either the capsule is seen to rupture, the total volume of 20 mL is injected or the patient fails to tolerate the procedure. Following the procedure all patients resumed physiotherapy treatment within 48 h.

As pain score data were categorical, chi-square test was used to assess the improvement in pain before and after the procedure with each group. We also considered that the difference in ROM before and after HD yielded data that were not normally distributed and thus, we used Wilcoxon signed rank test for paired samples.

Mann–Whitney test compared the improvement in the ROM between the groups. Statistical analysis was carried out using the IBM SPSS version 26; a two-tailed p value of less than 0.05 was considered statistically significant.

Results

The mean age at intervention was 59 years (range: 21 to 73 years). There were 168 females and 82 males. All diabetic patients in this series belonged to the idiopathic group. At three months following HD, pain scores and ROM of 250 patients were available. At 12 months, ROM data of 204 patients were available as 12 patients lost to follow-up and 34 patients required other treatments. The details are shown in Table 1.

Table 1.

Patients’ demographics based on aetiology and diabetic status.

FS type	3 months (total 250 patients)		12 months (total 204 patients)
	Non-diabetic	Diabetic	Non-diabetic	Diabetic
Primary	180	27	155	7
Post-traumatic	23	0	23	0
Post-surgical	20	0	19	0

Prior to injection, all patients had moderate or severe pain. Following HD, pain improvement was noted amongst the majority of patients. We calculated the number of patients with each category of pain before and after HD. We used chi-square test to assess the significance of the difference. Results of pain improvement are summarised in Table 2.

Table 2.

Pain improvement at three months after injection.

Pain category	Pre-injection Number of patients (%)	Post-injection Number of patients (%) (3 months follow-up)	p value*
No pain	0 (0%)	120 (48%)	0.0001
Mild	0 (0%)	110 (44%)	0.0001
Moderate	124 (49.6%)	20 (8%)	0.0001
Severe	126 (50.4%)	0 (0%)	0.0001

*Chi-square test.

ROM data were available for all patients at 3 months and for 222 patients at 12 months. The median of difference in flexion, abduction and external rotation before and after HD was calculated at 3 and 12 months. Using the Wilcoxon rank test, results showed that all parts of ROM have improved significantly following the HD at 3 and 12 months, which are shown in Table 3. In addition, the differences in all ROM pre- and post-HD were presented in box plots (Figures 1, 2 and 3).

Table 3.

Differences between the medians before and after hydrodilatation at 3 and 12 months.

	Pre-HD	3 months post-HD	12 months post-HD	p value (3 months follow-up)*	p value (12 months follow-up)*
All groups
Median flexion	50	120	110	0.01	0.01
Median abduction	50	110	100	0.01	0.01
Median external rotation	20	50	50	0.01	0.01
Idiopathic
Median flexion	50	120	110	0.03	0.006
Median abduction	50	110	110	0.0003	0.0001
Median external rotation	20	50	50	0.00001	0.00001
Post-traumatic
Median flexion	50	120	110	0.00001	0.00001
Median abduction	40	120	110	0.0007	0.006
Median external rotation	10	60	50	0.006	0.004
Postoperative
Median flexion	85	130	130	0.0001	0.0001
Median abduction	90	120	120	0.0001	0.0001
Median external rotation	20	50	60	0.004	0.006

*Wilcoxon rank-sum test.

Figure 1.

Improvement in external rotation at 3 months (blue) and 12 months (orange) following hydrodilatation.

Figure 2.

Improvement in flexion at 3 months (blue) and 12 months (orange) following hydrodilatation.

Figure 3.

Improvement in abduction at 3 months (blue) and 12 months (orange) following hydrodilatation.

When we compared the improvement in ROM based on the aetiology of FS, we found that the degree of improvement was not different between post-traumatic and post-surgical (p = 0.4), post-traumatic and idiopathic (p = 0.08) and between post-operative and idiopathic (0.59) groups. p values were calculated using the Mann–Whitney test.

In total, 34 patients (13.6%) needed further interventions within 12 months follow-up. Of these, 33 patients belonged to idiopathic group and 1 patient to post-operative group. Eight patients (3.2%) required repeat HD and 26 patients (10.4%) underwent arthroscopic capsular release (after failed repeated HD).

Diabetic patients were found to require more interventions following HD compared to non-diabetic patients: 20 out of 27 diabetic patients (73%) and 14 out of 223 non-diabetic patients (6.2%). This difference was found to be statistically significant p < 0.00001 (chi-square test). If we exclude diabetic patients from analysis, HD was successful in managing idiopathic FS, such that no further intervention was required, in 94% of cases irrespective of aetiology.

As diabetic patients required more interventions by 12 months since the procedure, seven patients remained available for assessment of the ROM. When we compared the improvement in ROM between diabetic and non-diabetic patients in the idiopathic group, we found no difference in the improvement in abduction at 3 months (p = 0.12) and 12 months (p = 0.6), in flexion at 3 months (p = 0.33) and 12 months (p = 0.9) and in external rotation at 3 months (p = 0.3) and 12 months (p = 0.9). Diabetic patients who responded to HD had comparable improvement in the ROM to non-diabetic patients.

Discussion

Despite being self-limiting, adhesive capsulitis often warrants treatment given the pain and severe functional restriction. ¹⁹

Introduced in 1965 by Andrén and Lundberg, ¹⁷ HD offers pain relief mainly through two mechanisms of action: the anti-inflammatory effects of steroid and the capsular rupture.^17,20 Rizk et al. believed that capsular rupture was the main reason for pain relief due to the decrease of stimulus on capsule pain receptor. ²¹ This is explained by the initial pain felt by patients at the start of the procedure followed by pain relief when the capsule is ruptured. In our study, we did not aim for capsular rupture and adopted a gradual stretching of the capsule.

Tveitå et al. ²² conducted a randomised controlled trial in 2008 on 76 patients with FS comparing steroid injection with HD to steroid injection alone. They found that the mean of improvement in Shoulder Pain and Disability Index from these two interventions was well below the minimal clinically important difference and that the two treatments were comparable. Patients in their series were assessed immediately after intervention and then at six weeks. Physiotherapy treatment was reported in some of patients before intervention and was optional afterwards. In our series, patients immediately restarted physiotherapy treatment to ensure sustainability of the capsular stretching that was achieved during HD. In our study, 92% of the patients had significant improvement in pain at 3 months and ROM at 3 and 12 months following HD.

Ajda Bal et al. ²³ compared the effectiveness of steroid injection with exercise against exercise alone using Shoulder Pain and Disability Index, University of California-Los Angeles end result scores and night pain and shoulder passive ROM as the outcome measures; they found all the mentioned parameters were improved in the second week; nevertheless, 12 weeks follow-up showed no significant change between the two groups. Unlike our study, which showed the improvement kept happening in all study groups in 3 and 12 months follow-up including diabetic patients.

Yoon et al. compared three types of injections used to treat FS: subacromial, intra-articular and HD on 29, 29 and 28 patients, respectively; they found that HD was superior in 3 months follow-up; however, at 12 months the pain and ROM were the same for all groups. ¹⁸ We believe that the management of FS is to provide early symptomatic relief. Given that it is a self-limiting condition most patients will have similar long-term outcomes irrespective of intervention.

Hamdan and Al-Essa ²⁴ compared HD combined with MUA versus MUA alone in the treatment of FS; the former was found to be superior compared to the latter. MUA, however, would require general anaesthesia and it could be associated with risks of humeral fractures or rotator cuff injuries.

Trehan et al. ²⁵ investigated the use of two consecutive HD injections with a six-week interval. They compared the shoulder scores of 22 patients who had a repeat HD to those of 16 patients who had single HD and found that the results were similar. In our series, we reserved repeat HD for patients with recurrence of symptoms and found that in the vast majority of cases a single intervention was sufficient.

Conclusion

Glenohumeral joint volume HD is a successful low risk intervention in the majority of patients irrespective of aetiology. Diabetic patients had more recurrence and needed further interventions after HD when compared to non-diabetic patients. However, HD still offered diabetic patients significant pain relief and improvement in the ROM at three months and if successful conferred similar improvement to the non-diabetic cohort.