Recurrence of the frozen shoulder after hydrodilatation, what is the true incidence?

Abstract

Purpose of the study

To assess the outcome and recurrence rates of frozen shoulder treated by hydrodilatation in an independent hospital setting.

Method

Patients presenting to a shoulder clinic from August 2019 to July 2021 with a diagnosis of frozen shoulder were offered hydrodilatation. Data included primary or secondary frozen shoulder, length of symptoms, and diabetic status. An Oxford Shoulder Score was completed prior to hydrodilatation. Using ultrasound guidance, 40 mg Triamcinolone and local anaesthetic (10–25 mL depending on patient tolerance) were injected into the rotator interval. At a mean of 9 months, patients recorded their tolerance of the procedure, subsequent progress, the need for further treatment, and their current Oxford Shoulder Score.

Results

From 55 shoulders, six patients had a failure to improve and 10 patients had a transient improvement followed by recurrence (29%). 2/21 (9.5%) patients had 25 mL injected compared to 14/34 (41%) who had < = 20 mL (p = 0.012). 14/43 (33%) of primary frozen shoulder patients had a recurrence, compared with 2/12 (16%) secondary frozen shoulder patients, p = 0.019.

Conclusion

Further treatment was indicated in 14/34 (41%) of patients who underwent hydrodilatation in the frozen stage of frozen shoulder and could not tolerate more than 20 mL of injection, and was more commonly required in primary (33%) versus secondary (16%) frozen shoulder.

Introduction

Frozen shoulder (FS) is a condition where the capsule of the shoulder joint becomes inflamed, leading to pain. This is followed by contraction of myofibroblasts within the shoulder capsule causing tightening of the capsule and a decrease in both the active and passive range of movements of the shoulder.^1–3 Plain X-rays show no abnormality but MRI may show inflammatory change with thickening and neovascularization at the rotator interval and coracohumeral ligament, (not always present but pathognomonic when seen) and obliteration of the fat triangle under the coracoid process. ⁴

The incidence of FS is estimated at 2–5%, but most notably higher (10–30%) in patients with diabetes mellitus. Other endocrine disorders such as thyroid disease are also known to be associated with higher incidences.^5–6 FS has been described in three phases; freezing, where the pain is the predominant feature, followed by stiffening; frozen, when the shoulder is stiff and painful, and thawing, when the pain diminishes and gradually the stiffness resolves. Typically, each of these stages lasts around 6 months although there is a variation of presentation and a much longer timescale in diabetic patients. ⁷ It is unknown why FS occurs or resolves but it can be categorised into primary (spontaneously occurring) or secondary (to a known precipitating factor, such as soft tissue injury to the shoulder, upper limb fracture, or cardiothoracic/breast surgery). ⁸ FS was first described by Codman in 1934 ⁹ and later refined by Lundberg in 1969, who first proposed a distinction between primary and secondary causes. ¹⁰ Treatment options include Non-Steroidal Anti-inflammatory Drugs, ¹¹ Cortisone injections,^11–15 Physiotherapy,^16,17 or combinations of non-operative treatments,^18–24 Manipulation under Anaesthetic (MUA),^10,25,26 Arthroscopic Release^27,28 and Hydrodilatation (HD).^29–32 HD involves the injection of Cortisone and saline under fluoroscopic or ultrasonic guidance into the glenohumeral joint.

All methods have an incidence of failure to improve and recurrence, but the reported rates of reintervention for HD (2–36%)^33–36 have been significantly lower than for MUA (15–40%).^26,37,38 However, studies showing a higher recurrence rate after MUA have been reported from an independent hospital setting compared to the low levels of reintervention following HD in the National Health Service (NHS) system. We, therefore, report our recurrence rate of FS after HD in an independent hospital setting.

Methods

All patients who presented to the senior author from August 2019 to July 2021 with a diagnosis of FS (pain and restricted range of movement in all planes with a normal plain X-ray) were offered MUA or HD with equipoise unless the patient had a contraindication to MUA (e.g. high risk for GA or known osteoporosis) in which case HD was offered. Data collected included age, sex, primary or secondary FS, length of symptoms, diabetic status and any comorbidities. An Oxford Shoulder Score (OSS) ³⁹ was completed prior to the procedure. HD was performed by one of two experienced musculoskeletal radiologists (AP & HT). Using Ultrasound guidance, the state of the rotator cuff was first assessed and recorded as normal, thinned, partial or full thickness tear, then 40 mg Triamcinolone and local anaesthetic (10–25 mL depending on patient tolerance) were injected specifically into the thickened rotator interval and coracohumeral ligament. ⁴⁰ Patients were given advice on mobilising the shoulder freely and all had an open appointment to return to the shoulder clinic if any concerns arose. At a mean of 9 months (range 1–28 months), all patients were sent a questionnaire (Appendix 1) which asked about the level of pain caused by the procedure (none, moderate or significant) and their current OSS. They were also asked if the procedure had relieved their pain, if so for how long or whether the improvement continued to date. Similarly, they were asked if the procedure had improved their range of movement and if so, for how long. Finally, they were asked if they had had any further procedures on their shoulder.

Results

Fifty-eight patients underwent HD, including one patient who had bilateral procedures. Two patients had undergone previous shoulder surgery, and two patients were either uncontactable or had an incomplete data set, so were excluded from the study leaving a total of 54 patients (55 shoulders). Patient demographics are shown in Table 1. Patients had reported symptoms for a mean of 8.4 months (IQR 6.5–11.3 months) prior to treatment. The ultrasound assessment prior to injection identified one patient with a known full-thickness supraspinatus tear, two patients with partial-thickness supraspinatus tear, and 52 shoulders with a rotator cuff within normal limits. The OSS of all patients’ pre- and post-HD is shown in Figure 1 alongside the subgroup of patients who underwent a further procedure. Sixteen patients (29%) had either insufficient improvement or recurrence of symptoms, and 15 of these patients requested further treatment. Of these 16 patients, six had a failure to improve with five seeking further treatment, and 10 patients had a transient improvement followed by recurrence (within a mean of 9 months from treatment). All 10 had further treatment (of the 15 patients, 11 chose MUA, and four repeated HD). This is shown in Figure 2. In addition, one patient had a cuff repair secondary to his HD for a known cuff tear, but neither patient with a partial thickness cuff tear required further treatment. 100% (two patients) with IDDM required further treatment (two MUAs). During HD all patients received 9 mL of 0.5% Chirocaine, 40 mg (1 mL) triamcinolone, and 0.9% saline to a total volume of between 10 and 25 mL. The amount injected depended on the tolerance of the patient to the injection. Figure 3 shows the number of shoulders injected with each volume and the incidence of recurrence. Three patients (5%) found the procedure painful, 14 (25%) moderately painful and 38 (70%) reported no pain. Only 2/21 (9.5%) patients who had a volume of 25 mL injected had a recurrence of symptoms, whereas recurrence occurred in 14/34 (41%) patients who had 20 mL or less injected. Using Fishers Exact test, there is a significant difference between these two groups (p = 0.012).

Table 1.

Patient demographics.

Age	58 (40–89)
Sex M:F	30:25
Non-diabetic	45
Non Insulin Dependent Diabetes Mellitus (NIDDM)	8
Insulin Dependent Diabetes Mellitus (IDDM)	2
Primary:Secondary FS	43:12

Figure 1.

The Oxford Shoulder Scores (OSSs) for all patients pre-procedure and post-procedure. A subgroup of the post-procedure OSS for patients who underwent a second procedure.

Figure 2.

Flowchart of patients in the study.

Figure 3.

A graph showing the number of shoulders that had a recurrence when grouped per volume of injection.

We compared the OSS and recurrence rates of primary versus secondary FS. There was no significant difference between the OSS for primary versus secondary FS before HD (28 vs. 28) or after HD (38 vs. 40). However, 14/43 (33%) of primary FS patients had a recurrence, compared with 2/12 (16%) secondary FS patients, which was significantly different (Fishers Exact test, p = 0.019). We had only eight NIDDM patients and two IDDM patients in the study, so insufficient to analyse statistically, but of note was that 2/8 NIDDM patients and 2/2 IDDM patients had a recurrence.

Discussion

Our results show that 29% of patients underwent either a failure to improve or a recurrence of FS. Those patients who had an injection of 25 mL had a significantly lower failure or recurrence rate (9.5%) compared to those injected with 20 mL or less (41%). A total of 33% of primary FS patients had a recurrence compared to 16% of secondary FS patients, although this result may be confounded by smaller injection volumes.

Factors which may influence the outcome or recurrence rate of HD include patient factors such as diabetic status, primary or secondary FS, stage of the disease (length of symptoms), tolerance to the discomfort of the procedure and volume at which the capsule ruptures. Injection factors include the concentration of Cortisone, volume and location of the injection. There are few studies in the literature on recurrence rates after HD. Bell et al. ³⁰ reported on 109 shoulders in 106 patients, 15 with diabetes. They injected 40 mg Triamcinolone, local anaesthetic and 10–55 mL saline to achieve capsular rupture. Patients were reviewed at two and four months. There were 29 repeat HD procedures performed (26%), 22/94 non-diabetics (23%) and 7/15 diabetics (47%). These results are similar to our study, but no symptom length, site of injection or correlation with injection volume is reported. Haughton et al. ³³ report a 2% recurrence rate after HD, but excluded 16% of patients due to pain intolerance before introducing the use of Entonox during the procedure, and had a very short follow-up period (mean of 3.5 months). Interestingly, they found a negative correlation between outcome (as measured by improvement in OSS) and volume of injection (10–100 mL). Nicholson et al. ³⁴ report a large study of 2432 cases followed up for 5.4 years and found a symptomatic recurrence rate of 7.6% (12.3% in diabetics vs. 5.9% non-diabetics). In their study, distension of 50 mL versus 30 mL showed a significantly decreased odds ratio for recurrence of 2.2 (95% CI: 1.6–3.0, p < 0.001). Capsular rupture (which occurred in 61% of their patients) or steroid dose did not affect the rate of recurrence. Patients with diabetes and bilateral symptoms were more at risk for symptomatic persistence. This study excluded 121/2814 due to a painful procedure, records their recurrence rate within 3 months, and patients had a minimum of 6 months of symptoms before treatment, although no mean or range is given. The waiting time for decision to perform HD and the procedure is not specified, although patients who returned at 3 months with recurrence had repeat HD at a mean of 9 months, suggesting a 6-month waiting time. We note that many of their patients were able to tolerate large volumes of injection (50 mL). A possible explanation is that if the patient was entering the thawing phase of FS then the capsule may be more distensible and the likelihood of ease of distension and reduction in recurrence could occur, or that the capsule ruptured, relieving the pressure. In our study, the range of length of symptoms prior to HD is 6.5–11.3 months, thus all our patients were in stage 2 of FS. Patients who have 10 mL injected probably have very little in the way of a distension effect, and the recurrence rate was 50% in the Nicholson study. ³⁴ Those having larger volumes may either have a greater distension effect or may already have a relaxed capsule as described above. The rupture of the capsule does not seem to alter the outcome. With regards to steroid dosage, the Nicholson study compared 40 mg with 80 mg Triamcinolone acetonide or methyl prednisolone acetate and found no difference in the recurrence rates (p = 0.275). ³⁴ One study concluded that anterior injection was superior to posterior. ⁴¹ One study found no difference between outcomes or recurrences in primary versus secondary FS. ⁴² Thus, it is hypothesized that the capsule may respond to the cortisone injection by a diminution of the inflammation, leading to reduce pain, and a relaxation of the capsule leading to a greater range of movement. However, once the effect of the cortisone has worn off (6–12 weeks) then whatever the trigger to FS is may still be present in some patients leading to a recurrence. We know that patients with IDDM have a more prolonged course of stiffness and pain than non-diabetic patients and this may explain why these patients have a higher chance of recurrence. In our study, all patients were at the frozen stage (pain and stiffness) when the HD was performed, and this may account for the lower volumes injected and the higher recurrence rates than other studies. Previous studies have shown that there may be a difference in outcome for any intervention between NHS (public) patients and patients within the independent sector due to factors such as co-morbidities, length of symptoms, patient expectation and access to care.^43–45 We know that all patients in our study had immediate access to return to our clinic and therefore are confident that our recurrence rate is a true reflection of the incidence of recurrence and not lowered by patients’ willingness to tolerate lack of improvement or recurrence, or their inability to re-access care in this event.

Recurrence rates after MUA in the independent sector are reported. Non-diabetic patients had a 15.5% incidence of repeat MUA within 1 year, 2.4% at 1–5 years and 1% at more than 5 years from initial MUA. ³⁸

We have found that, by using ultrasound-guided injections, targeting the thickened rotator interval, 25 mL appears to be a sufficient volume of injection to have an efficacious response, has a low recurrence rate, and is well tolerated by the patient without requiring additional analgesia. It has the additional advantage of being able to visualise the rotator cuff and diagnose any additional pathology at the same time. Higher rates or recurrence occur with volumes smaller than 25 mL.

The strengths of this study are unique reporting of HD in a UK-independent setting, using targeted USS guided injection of the rotator interval. Our data collection is robust and we have a very low loss to follow-up, thus we are confident that our assessment of recurrence rates is accurate. However, our numbers are small thus limiting any statistical analysis for subgroups. It is a single surgeon series and definitions of recurrence or need for further intervention are subjective. We are aware that recurrence may occur at any stage after the procedure, although has been shown to be most common within the first year after MUA, ³⁸ so our follow-up interval of a mean of 9 months may be too short to capture all recurrences which may subsequently occur.

Conclusion

Ultrasound-guided injection of the rotator interval using 25 mL total volume gives a low rate of recurrence of FS (9.5%) at a mean of 9 months and is well tolerated by patients within the freezing or frozen stages of FS without the need for additional analgesia. Smaller volumes of injection have resulted in significantly higher recurrence rates (41%). We would cautiously advocate the use of this technique for the treatment of FS, but long-term studies with larger patient numbers are required to confirm this.

Appendix 1