What volume of injection should be used in hydrodilatation for frozen shoulder? A prospective cohort study

Abstract

This study evaluated the outcome and recurrence rates after hydrodilatation (HD) in patients with frozen shoulder (FS), comparing the effects of injecting 25 mL or more of fluid versus lower volumes. A total of 132 shoulders (130 patients) were treated at a shoulder clinic between August 2019 and February 2023. HD was performed under ultrasound guidance, injecting a combination of 40 mg Triamcinolone, 10 mL of local anaesthetic, and saline to a total volume of 10 to 40 mL based on patient tolerance. The results were analysed based on the volume of fluid injected and whether patients were diabetic. The overall failure or recurrence rate (FRR) was 24% after a minimum of 1 year. Non-diabetic patients injected with ≥25 mL had a significantly lower FRR (9%) compared to those injected with <25 mL (30%, p = 0.006). Diabetic patients had a much higher FRR of 52%, compared to 17% in non-diabetics (p = 0.0002). The study concluded that injecting patients with ≥25 mL of fluid had a significantly lower FRR than those injected with <25 mL at a minimum of 12 months follow-up (9% vs 30%, p = 0.006). Diabetic patients experienced a higher rate of recurrence, suggesting the need for patient counselling about the increased likelihood of failure.

Introduction

Frozen shoulder (FS) was first described by Codman in 1934 as global restriction of shoulder movement with limitation of external rotation and elevation and normal shoulder radiographs. ¹ It has been classified as either primary (occurring spontaneously) or secondary, due to other precipitating factors such as a soft tissue injury to the shoulder, upper limb fractures or cardiothoracic/breast surgery.^2,3

FS arises as a result of inflammation of the shoulder capsule and contraction of myofibroblasts which result in pain and restriction of movement.^4–6 MRI may show thickening of the coracohumeral and middle glenohumeral ligaments. ⁷

The incidence of FS is around 2% to 5% in the UK population with an increased incidence in patients with diabetes mellitus (10–40%),^8,9 and in females when compared to males (up to 70%). ¹⁰

The natural history of FS is described in three stages: freezing, frozen and thawing (Stages 1–3). Each stage lasts approximately 6 months, longer in those with diabetes mellitus. ⁹ Patients with other endocrine pathologies such as thyroid disease are also known to have a higher incidence of FS. ¹¹

Treatment options for FS include simple analgesia, non-steroidal anti-inflammatory drugs, ¹² cortisone injections,^12–16 physiotherapy,^17–25 manipulation under anaesthesia (MUA),^2,26,27 arthroscopic capsular release (ACR)^28,29 or hydrodilatation (HD).^30–33

HD involves the injection of fluid (saline ± local anaesthetic ± cortisone ± contrast ± air) into the glenohumeral joint under ultrasound or fluoroscopic guidance. ³⁰

All methods have an incidence of failure and recurrence particularly in those patients with diabetes mellitus, and although there are many reports of the success of these various techniques, relatively little has been published on recurrence rates.^{12,20,27,31,34,35} In one large study however, it has been shown that the volume of fluid injected in HD has been correlated with reintervention rates, with larger volumes (>30 mL) being associated with the lowest rates of reintervention. ³⁶

In our pilot study, those patients who did receive 25 mL or more had a lower rate of failure or recurrence than those receiving smaller volumes (20 mL or less). Using Fishers exact test, this difference (5% vs 35%, respectively) was significant (p = 0.008). ³⁷ We therefore extended our study to see how those patients who could tolerate 25 mL or more compared to those who tolerated 24 mL or less both in outcomes and failure or recurrence rate (FRR) and present our findings. In addition, we compared diabetic patients with non-diabetic patients, as previous studies using both HD and MUA had shown increased FRR in diabetic patients.^27,31,34,38

Methods

All patients presenting to the senior author with FS (as evidenced by pain and a restricted range of active and passive movement of the shoulder in any plane, with no clinical evidence of cuff pathology and a normal plain radiograph), between August 2019 and February 2023, were offered HD. Patients who had rheumatoid arthritis, previous shoulder HD, MUA or surgery, and those with a known rotator cuff tear were excluded from the study.

The procedure was performed under ultrasound guidance by one of two experienced musculoskeletal radiologists (AP and HT). The integrity of the rotator cuff was assessed prior to the HD taking place, and the HD was not carried out if the rotator cuff was torn. Forty milligrams triamcinolone, 10 mL of 0.5% chirocaine and variable amounts of saline up to a maximum total volume of 40 mL were injected into the rotator interval and coracohumeral ligament. The volume of injection depended on the patient's ability to tolerate the HD. The shoulder capsule was not ruptured in any of the patients.

Data collected included the patient's age, sex, duration of symptoms (DOS), diabetic status, total volume of fluid injected, and rotator cuff status. An Oxford Shoulder Score (OSS) ³⁹ was obtained prior to HD in all patients. Following the procedure, patients were encouraged to mobilise their shoulder without restrictions, but no formal outpatient physiotherapy was requested. At a minimum of 12 months from HD (mean 13.5 months, range 12–15 months) patients were followed up by postal or telephone questionnaire (Appendix 1) which asked about the level of pain induced by the HD (none, moderate, severe), their current OSS, to what degree the HD relieved their pain, if their range of movement (ROM) had improved and if they had required any further treatment. We chose this time interval because in a study of outcome after MUA for FS, recurrences were uncommon beyond 1 year from treatment. ⁴⁰

Failure to improve (FTI) was recorded if the patient reported subjectively inadequate improvement at any stage after HD. Recurrence was recorded if there was a deterioration in the patient's pain or function after an initial improvement.

Patients were separated into three groups: group 1 (those injected with <25 mL fluid), group 2 (those injected with ≥25 mL fluid) and group 3 (diabetic patients with any volume of injection). These groups were compared for pre-HD OSS and DOS, as these were factors which might affect the outcome. Statistical analysis of our results was performed using SPSS^© version 28.

Results

A total of 145 shoulders (143 patients) were referred for HD. Six patients had undergone previous shoulder surgery, and 2 patients had an incomplete data set. The ultrasound assessment prior to injection identified 4 shoulders with a full thickness supraspinatus tear and 1 shoulder with a subscapularis tear. These patients were therefore excluded from the study. Two shoulders with a partial thickness supraspinatus tear, 4 shoulders with mild calcification or tendinopathy and 126 shoulders in which the rotator cuff was reported as within normal limits were all included in the study. A total of 132 shoulders underwent HD including two patients who had bilateral procedures (130 patients in total). Patient demographics are shown in Table 1.

Table 1.

Patient demographics.

	Group 1 (<25 mL)	Group 2 (≥25 mL)	Group 3 (diabetic)
No. of shoulders	40	65	27
Sex M:F	18:22	27:38	18:9
Mean age	57 (45–64)	55 (40–63)	42 (31–50)
Mean duration of symptoms (weeks)	14 (8–21)	16 (9–23)	14 (6–22)
Mean pre-HD OSS	28 (24–33)	29 (24–34)	26 (20–31)

Interquartile ranges (IQR) in brackets.

Comparing group 1 with group 2, the ratio of male to female patients is not significantly different using Chi-squared (Χ²) analysis (p = 0.72). Comparing groups 1 and 2 (all non-diabetic patients) with group 3 (diabetic patients) by the same method, the ratio of male to female patients is significantly different (p = 0.01), with a male predominance in the diabetic group.

Using Mann–Whitney U (MWU) tests, there is no significant difference in mean age (p = 0.144), DOS (p = 0.194) and pre-HD OSS (p = 0.197) between groups 1 and 2. There is a significant difference between groups 1 and 2 (non-diabetic) compared with group 3 (diabetic) for mean age, and pre-HD OSS (p < 0.001 and p = 0.01, respectively). There was no significant difference in DOS between these groups (p = 0.435).

In the non-diabetic patients, 40/105 (38%) had <25 mL injected and 65/105 (62%) had ≥25 mL injected. This was a similar ratio of volumes injected seen in the diabetic group; 10/27 < 25 mL (37%) and 17/27 ≥ 25 mL (63%). X² analysis demonstrated no significant difference in the injection volume ratios between the two groups (p = 0.919).

At follow-up (FU) (mean 13.5, range 12–14 months), mean OSS was 42 (IQR 34–45) in group 1, 44 (IQR 42–46) in group 2 and 39 (IQR 33–45) in group 3. The mean improvement in OSS was therefore 14, 15 and 13, respectively, there was no significant difference between the groups (MWU; p = 0.27). No patients were lost to follow-up. These results are shown in Figure 1.

Figure 1.

Pre- and post-HD OSS scores.

In group 1, (<25 mL, 40 shoulders), 4 shoulders failed to improve and 8 had recurrence of which 9 had repeat HD, 2 had MUA and 1 had no further treatment. In group 2, (>25 mL, 65 shoulders), 1 shoulder failed to improve and 5 had recurrence, of which 3 had repeat HD, 2 had MUA and 1 had no further treatment. In group 3 (diabetic, 27 shoulders), 5 failed to improve and 9 had recurrence which 4 had repeat HD, 8 had MUA and 2 had no further treatment.

Using X² analysis, there was a significant difference in the FRR between group 1 (<25 mL) and group 2 (≥25 mL) (p = 0.006). Group 1 had 12/40 FRR (30%) and group 2 had 6/65 FRR (9%). There was also a significant difference in the FRR between non-diabetics (18/105, 17%) and diabetics (14/27, 52%) (p = 0.0002). These results are presented in Figure 2.

Figure 2.

Failure and recurrence rates (FRR).

Further statistical analysis using the Spearman Rank Correlation assessed the impact of each variable on FRR. In the non-diabetic groups there is no statistically significant correlation between age, DOS or pre-HD OSS and FRR in the non-diabetic groups (p = 0.809, p = 0.514, p = 0.142, respectively). In the diabetic group, there were statistically significant weak negative correlations between both age and pre-HD OSS with FRR (p = 0.023 and p = 0.029, respectively). There was no significant correlation between DOS and recurrence (p = 0.228).

X² analysis demonstrated no significant difference between gender and recurrence in groups 1 and 2 (p = 0.086). There was also no significant difference between gender and recurrence in group 3 (p = 0.173). However, comparing the non-diabetic and diabetic groups, we saw a statistically significant difference between gender and recurrence (p = 0.011) with a male predominance. It is important to note that the diabetic group is predominantly male, which has a higher recurrence rate as previously outlined.

Of the 132 shoulders injected, 7 patients reported severe pain, 17 reported moderate pain and 108 reported mild/no pain. Of the 7 patients reporting severe pain, 5 (71%) subsequently had a FTI, significantly more than those who had no, mild or moderate pain (4/125, 3%) (p < 0.001). Six of these patients had <25 mL injected and 1 had ≥25 mL injected.

Discussion

An analysis of the results reveals a 30% FRR in patients receiving <25 mL, a 9% FRR in patients receiving ≥25 mL and a 52% FRR in diabetic patients. We established that gender, age, DOS and pre-HD OSS were not confounding factors; however, diabetic status did significantly increase the risk of recurrence, as did a younger age within this group.

There are other factors that may influence the recorded outcome and FRR after HD. These include the stage of FS (freezing, frozen, or thawing), the inherent tightness of the shoulder capsule, the tolerance of the patient to the procedure, whether or not the shoulder capsule ruptures, the dose of steroid used, the location of the injection, and the length of follow-up. In addition, ease or willingness to return for further treatment may alter the recorded recurrence rate, as will patients lost to follow-up.

FS is a self-limiting condition therefore most patients’ symptoms will resolve over time without treatment. We know that our patients are all in the first or second stage of FS, so both their treatment and FU at 1 year are still within the expected timescale for the presence of the condition untreated. We therefore would not have expected them to have recovered in this timescale without treatment. Whatever factor causes or maintains FS during this time is potentially still active, hence the risk of FRR after treatment. Previous studies on outcomes of HD and recurrence rates often do not specify the DOS nor the stage of FS, but it is conceivable that patients in the thawing stage might tolerate higher volumes and be less likely to have a recurrence. In our study, there was no significant difference in the DOS between groups, and no correlation between DOS and FRR.

Vad et al. concluded that results of HD are related to stage of FS at time of treatment, but this is a small underpowered study. ⁴¹ In a study using MUA, no difference in outcome was seen in patients with varying symptom duration, but all were within FS stage 1 or 2. ⁴²

The tightness of the capsule may be a static (capsular volume) or dynamic (capsular resistance to dilatation) factor. Patients with capsular tightness have a decreased ROM and may be less able to tolerate higher volumes of injection or may require higher pressure of injection which might precipitate capsular rupture. Our study did not record ROM, and no patient had a capsular rupture because the injection was only performed up until the volume tolerated by the patient. Clearly there is variation in the patients’ tolerance of larger volumes. The normal shoulder capsule volume has not been defined but we speculate that studies recording injection of >30 mL are either rupturing the capsule or injecting an already distended or distensible capsule, which may well determine the outcome. Pimenta reported a mean of 25 mL injection volume in their capsule preserved group and a mean of 36 mL in their capsule ruptured group. ⁴³

The severity of the disease may be a factor in outcome or FRR. The OSS is one measure of this. In our study, there was no significant difference in the pre-HD OSS between groups 1 and 2, but there was a lower mean pre-HD OSS in group 3 (diabetics) compared to groups 1 and 2 together (non-diabetics). Diabetes may predispose to a more severe form of FS hence a higher FRR.

HD was a painful procedure in some of our patients. In our study, only 7 patients reported severe pain but 5 of these had FTI. It may therefore be necessary in some patients to provide some form of analgesia before or during the procedure in order both to keep the patient comfortable, and to allow a greater volume to be injected. Previous studies have drawn conflicting conclusions on whether capsular rupture is beneficial.^44,45 Although higher volumes may result in more capsular distension, a rupture may occur before all parts of the capsule are distended and may disperse the steroid extra-articularly. Studies showing a decreased success rate (although not necessarily increased recurrence rate) with larger injection volumes may be confounded by differences in DOS and stage of FS at time of treatment. Bell et al. reported their results with a data set of 109 shoulders. They concluded that maximum distention is required to cause capsular rupture to ensure effective treatment. They injected 1 mL of betamethasone, 2 mL of 2% lidocaine and 10 to 55 mL of normal saline to achieve capsular rupture. Their overall recurrence rate was 23% in their non-diabetic patients. ³¹

Nicholson et al. in 2020 reported on their large data set of 2432 cases collected over 5 years. 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 details of the procedure are not specified. Their overall reintervention rate was reported at 7.6% with lower reintervention rates reported in those patients who had greater volumes injected (>30 mL). Of note was that patients injected with only 10 mL had a 50% reintervention rate. ³⁶ Both studies concluded that capsular distention was the key to successful long term patient outcomes and lower recurrence rates.

The majority of published papers on HD for FS have used steroids. Bell injected 40 mg triamcinolone. ³¹ Nicholson compared 40 mg triamcinolone with 80 mg triamcinolone or methylprednisolone and found no difference in recurrence rates. ³⁶ We have used 40 mg triamcinolone in our study.

A meta-analysis on the site of injection in HD has concluded that the rotator interval approach gives improved pain relief at 12 weeks compared to a posterior approach, but there is no data on comparison of failure or recurrence rates. ⁴⁶ In our study we have used the rotator interval approach.

There is a wealth of literature on the comparison of diabetic versus non-diabetic patients with FS using each of the previously described methods of treatment. Many have shown a higher recurrence rate in diabetics, suggesting that this is due to some aspect of the natural history of the disease rather than the method used to treat it.^31,34,47 Jenkins et al. published a series of 315 shoulders with FS, 39 shoulders in the diabetic group. The diabetic patients were at a 36% increased risk of requiring a further MUA, compared with the 15% increased risk in the non-diabetic group. There was also a higher recurrence rate in those patients who were insulin dependent compared to non-insulin dependent diabetics. ³⁴ Using HD, Bell had a recurrence rate of 47% in their diabetic patients and 23% in their non-diabetic patients. ³¹ Nicholson reported that the requirement for further intervention for persistent symptoms following arthrography was significantly associated with Diabetes (p < 0.001), ³⁶ in keeping with our findings, and justifying our decision to analyse the results in diabetics separately.

Our study has chosen to assess the recurrence rate at 1 year because, in a study of outcome after MUA for FS, recurrences were rare beyond 1 year from treatment. ⁴⁰ We had no patients lost to follow-up and are therefore confident that our figures represent a true reflection of the failure or recurrence rates following HD with these varying volumes.

The use of giving information on rehabilitation to the patient rather than a formal physiotherapy appointment was based on the evidence from a study of a comparison of these two options following MUA for FS, which showed no clinically important difference in outcome. ⁴⁸

This study had some weaknesses. It is not a randomised controlled trial, so patients were divided into <25 mL or ≥25 mL injections based on their tolerance of the procedure rather than randomly.

The number of patients is relatively small and therefore may cause a type 2 error. ROM was not recorded pre- or post-HD, and this might have correlated with the volume injected or the risk of failure or recurrence.

Conclusion

Ultrasound guided HD using a total volume of 25 mL or more gives a significantly lower FRR of FS at a mean of 12 months follow-up in comparison with smaller volumes (9% vs 30%, p = 0.006). Diabetic patients have a significantly higher recurrence rate then non-diabetics (52% vs 17%, respectively, p = 0.002). We would propose a high volume of fluid, at least 25 mL, to enable capsular distention in order to reduce the incidence of recurrence. Some patients may require analgesia at the time of injection to achieve these volumes. Diabetic patients presenting with FS should be counselled about their higher failure and recurrence rates, and therefore the increased risk of the requirement for further treatment.