Optimising the use of physiotherapy resources after manipulation under anaesthetic for frozen shoulder

Abstract

Introduction

Manipulation under anaesthetic (MUA) is a successful treatment for frozen shoulder (FS), and the recovery period and recurrence rates may be reduced by postoperative physiotherapy. This study evaluates two physiotherapy pathways for patients undergoing MUA for FS.

Methods

Between 2016 and 2018, 248 age- and sex-matched patients presented to either a NHS secondary care upper limb service or the lead author’s independent practice with a diagnosis of FS. The patients had differential access to postprocedure physiotherapy based on which service they presented to. In Group 1, physiotherapy advice only was given to the patient. In Group 2, supervised hydrotherapy and physiotherapy occurred postoperatively. Pre- and postprocedure Oxford Shoulder Scores (OSS) were collected for each group. Analysis of covariance (ANCOVA) was used to measure the effect of physiotherapy on postoperative OSS.

Results

Group 2 showed a significantly greater improvement in postprocedure OSS when compared with Group 1 (18.2 vs 16.7) p<0.001). The estimated maximum effect of physiotherapy on postoperative OSS was an increase of 3.2.

Conclusion

Following MUA for FS, a statistically significant increase in OSS was detected in patients receiving postprocedure physiotherapy compared with advice alone. There was no difference in recurrence rates. The increase in OSS (3.2) is below the minimal clinically important difference, raising questions regarding the relative importance of postprocedure physiotherapy in a resource-limited environment.

Introduction

Frozen shoulder (FS) is a well-described condition of the shoulder in which pain and stiffness occurs.¹ Although in the majority of cases spontaneous resolution occurs in months to years,^2,3 patients with this condition often have severe pain and disturbed sleep, causing inability to perform activities of daily living. The British Elbow and Shoulder Society have produced a pathway for the management of FS in which physiotherapy, injections of cortisone, manipulation under anaesthetic (MUA), arthroscopic release or hydrodilation are suggested.⁴

MUA can significantly improve both pain and range of movement (ROM) within 3 weeks.^4–8 There is, however, a failure rate of the procedure of 4–7% within 3 months, and a later recurrence rate of 10–40%, which is related to diabetic status.⁹ The recurrence rates for hydrodilation have been reported as low as 1%, but this may be due to other factors not considered, such as access to a repeat procedure.¹⁰ Failure and recurrence rates for arthroscopic release have not been reported to the best of our knowledge.

Physiotherapy following the procedures has been advocated as a way of maximising the ROM, relieving pain and potentially reducing the failure or recurrence rate.¹¹ However, it is unclear whether physiotherapy following surgical intervention alters outcomes.

This study compares two pathways for patients undergoing MUA for FS: one in a NHS setting where physiotherapy advice only is given to the patient, and the other in an independent hospital setting where supervised hydrotherapy and physiotherapy occur postoperatively.

Methods

Between 2016 and 2018, 295 patients presented to either a NHS secondary care upper limb service, or the lead author’s independent practice with a diagnosis of FS. All patients were informed of the diagnosis, its natural history and the treatment options available, which were MUA, or hydrodilation if the patient was not suitable for MUA.

A total of 248 age- and sex-matched patients underwent MUA and were then divided into two groups based on the access to postprocedure physiotherapy at the institution they presented to: Group 1 were given physiotherapy advice with no face-to-face contact; Group 2 were given supervised postprocedure hydrotherapy and physiotherapy.

Inclusion criteria

Group 1 (MUA only)

All patients aged 16–80 years presenting to a NHS upper limb clinic with a diagnosis of primary or secondary FS who were seen and treated by MUA between November 2016 and September 2018 were included. This group were given written information only on mobilisation of their shoulder postoperatively. Patients had their MUA performed at a mean of 20 weeks (range 2–34 weeks) from the clinic appointment.

Group 2 (MUA plus physiotherapy)

All patients aged 16–80 years presenting to the independent clinic of the senior author with a diagnosis of primary or secondary FS treated by MUA between November 2016 and September 2018 were included. This group had supervised hydrotherapy and physiotherapy postoperatively. Patients had their MUA performed at the first opportunity with a mean wait of 3 weeks (range 1–8 weeks).

Exclusion criteria

Patients with a proven rotator cuff tear on magnetic resonance imaging (MRI) or ultrasound scan, any form of glenohumeral arthritis on x-ray, or who had had a previous MUA on that shoulder were excluded from the study, as were patients who underwent hydrodilation (due to high risk of complication from general anaesthetic (GA), patients who wished to avoid a GA or in those with radiologically apparent osteoporosis).

The primary outcome measure was the Oxford Shoulder Score (OSS).¹¹ The secondary outcome was the need for redoing MUA or further treatment. All patients underwent a standardised MUA, performed as a day case procedure. On the day of the procedure, they completed an OSS. The patient underwent a GA in the supine position. Holding the patients arm between the shoulder and the elbow, and fixing the scapula with a hand over the acromion, the ROM of flexion, abduction and internal and external rotation was assessed by eye to the nearest 5° and recorded. The shoulder was then manipulated into abduction followed by flexion, external rotation, cross-body adduction and internal rotation. MUA was repeated until the maximum ROM had been achieved, and the final ROM in these planes was recorded. Then 10ml 0.5% chirocaine, and 80mg Depo-medrone (Pharmacia) was injected into the glenohumeral joint before GA was ceased.

Postoperatively, patients in Group 1 were given written physiotherapy instructions (Appendix 1). They were followed up by a senior physiotherapist by telephone at an average of 4 weeks, where a further OSS was recorded along with questions on compliance with physiotherapy advice, satisfaction with the procedure or request for further treatment. Outcome options were either discharge, referral for outpatient physiotherapy or return to see the surgeon in the clinic.

Patients in Group 2 underwent the same MUA but had hydrotherapy within 48h of the procedure and a first physiotherapy session within 5 days of the procedure. In cases where patients had regained a full/near-full ROM and were happy with progression they did not have further physiotherapy sessions, but were encouraged to continue with their exercises until reviewed in clinic. However, if still in discomfort or lacking ROM, then physiotherapy continued at the discretion of the physiotherapist and the patient. In some cases, this continued for several weeks after the manipulation. Follow-up was conducted by the senior author within 4 weeks of the procedure, where a current OSS was recorded, and outcome options were discharge, or further physiotherapy. If further physiotherapy was unsuccessful, then repeat MUA was offered.

Statistical analysis

The statistical analyses were performed in R version 3.6.1 (R Core Team). Analysis of covariance (ANCOVA) was used to analyse differences between the groups as it is expected to produce less-biased estimates of effect size than alternative methods. Fishers exact test was used to analyse recurrence rates between the two groups.

Sample size

Studies of the OSS reported standard deviation (SD) and minimal clinically important difference (MCID) of 9 and 6 points, respectively.^12,13 Equal group sizes of 49 were calculated to provide 90% power to detect the MCID at the 0.05 level of significance in a two-sample t-test. As we had over 100 patients in each group, we concluded that our study was adequately powered to test our hypothesis (see Results).

Matching

Exact matching on sex, diabetes disease status and primary versus secondary procedure was performed as these are factors that have been shown previously to have an effect on outcomes.⁷ Exact matches were found for 129 (91%) MUA-only patients and for 119 (99%) MUA plus physiotherapy patients.

We did not match for factors that have previously been shown to have no effect on outcomes, namely length of time of symptoms prior to MUA,⁸ age or initial OSS⁷ (the latter is also accounted for in the ANCOVA analysis).

Results

A total of 248 age- and sex-matched patients were available for analysis. Of these 248 patients, 129 underwent MUA followed by physiotherapy advice alone (Group 1) and 119 underwent MUA followed by supervised physiotherapy (Group 2). Four patients in Group 1 and six patients in Group 2 chose hydrodilation and were excluded. Five patients in Group 2 did not attend postprocedure physiotherapy and were therefore excluded. All patients included and analysed are detailed in Figure 1 and Table 1.

Figure 1 .

Recruitment and number of analysed patients. MUE = manipulation under anaesthetic

Table 1 .

Group characteristics

		MUA only (n=129)	MUA+physiotherapy (n=119)
Sex	Male	58 (38%)	59 (50%)
	Female	71 (62%)	60 (50%)
Diabetes	None	94 (73%)	101 (85%)
	Type 1	11 (9%)	4 (3%)
	Type 2	24 (19%)	14 (12%)
Classification of FS	1°	84 (65%)	88 (74%)
	2°	54 (45%)	31 (26%)

FS = frozen shoulder; MUE = manipulation under anaesthetic

Oxford shoulder scores

Mean ± interquartile range (IQR) preoperative OSS was 24.6±6.8 (23.0±5.1 for MUA only and 26.4±8.0 for MUA plus physiotherapy).

Mean postoperative OSS was 42.0±7.6 (39.8±8.4 MUA only and 44.6±5.6 MUA plus physiotherapy).

Mean improvement in OSS was therefore 17.4±8.0 (16.7±8.0 MUA only and 18.2±8.1 MUA plus physiotherapy). Given variation in OSS in the two groups before treatment and the difference in improvement in OSS after treatment, ANCOVA analysis shows the maximum difference in OSS caused by the differing physio regimes is 3.2. Scores are shown in Figure 2.

Figure 2 .

Pre- and postoperative OSS. OSS = Oxford Shoulder Score

Range of movement

ROM data were available for 101/129 patients in Group 1 and 119/119 patients in Group 2. Mean ROM (±interquartile range) is listed in Table 2.

Table 2 .

Mean ROM (degrees)

		Pre-MUA	Post-MUA
MUA+advice (Group 1)	Flexion	70 (65–75)	170 (165–175)
	Abduction	70 (65–75)	170 (165–175)
	ER	20 (15–25)	60 (55–65)
	IR	30 (25–35)	60 (55–65)
MUA+physiotherapy (Group 2)	Flexion	70 (65–75)	170 (165–175)
	Abduction	70 (65–75)	170 (165–175)
	ER	20 (15–25)	60 (55–65)
	IR	30 (25–35)	60 (55–65)

ER = external rotation, IR = internal rotation; MUE = manipulation under anaesthetic; ROM = range of movement

Recurrence rate

Analysis of the recurrence rate using Fisher's exact test showed that there was no statistically significant difference between the two groups in the number of repeat MUAs, either early (<3 months) or late (>3 months) (Table 3).

Table 3 .

Comparison of rates of repeat (Redo) MUA between matched patients in the study groups

	RedoMUA<3 months	RedoMUA>3 months	Total Redo MUA
MUA only (129)	2	11	13
MUA plus (119) physiotherapy	5	13	18
p values (Fishers exact test)	0.25	0.39	0.18

MUE = manipulation under anaesthetic

A comparison of the mean preoperative (preop) OSS and preop and postoperative (postop) ROM between those patients who had a recurrence, compared with those in the same group who did not, was performed. The mean (±IQR) preop OSS was 22.8(±9.4) vs 23.4(±5.0) (p=0.43) for those patients in Group 1 (MUA only) who did (n=13) vs did not (n=116) have a recurrence. In Group 2 (MUA plus physio) the mean preop OSS was 26.2(±7.8) vs 25.9(±10.1) (p=0.41) for those patients who did (n=18) vs did not (n=101) have a recurrence. Analysis of the ROM of the patients with recurrence compared with those without recurrence in each group similarly revealed no difference in the preop ROM (p=0.45 in Group 1 and p=0.41 in Group 2).

The recurrence rate was greater for diabetic compared with nondiabetic patients in both groups as listed in Table 4.

Table 4 .

Recurrence rates in patients with DM

	Group 1 n=129	Group 1 recurrence n=13	Group 2 n=119	Group 2 recurrence n=18
Type 1 DM	11 (9%)	2 (15%)	4 (3%)	1 (6%)
Type 2 DM	24 (19%)	4 (31%)	14 (12%)	4 (22%)
Nondiabetics	94 (72%)	7 (54%)	101 (85%)	13 (72%)

DM = diabetes mellitus.

Discussion

This study reports a statistically significant improvement in OSS as measured at 4 weeks postoperatively with postprocedure supervised physiotherapy compared with advice alone. The difference in OSS of 3.2 was below the MCID of 6. Both groups made significant improvements in OSS post-treatment (improvement in OSS 18.2 and 16.7, respectively). There was no difference in the recurrence rate with either treatment (p=0.18). Diabetic patients had double the risk of recurrence compared with nondiabetics, but this was similar in both groups so supervised physiotherapy does not appear to reduce this recurrence rate. No patient characteristics were identified that may predict benefit from supervised postprocedure physiotherapy.

MUA has been shown to be an effective treatment for FS,^4–9 but it has a failure rate (defined as no improvement in symptoms or ROM at any time following MUA) reported as 4%, and a recurrence rate (defined as improvement after MUA followed by deterioration of pain or ROM at any time in the future) of 11–40%.⁹ Factors affecting the outcome have been shown to be gender and diagnosis of primary (idiopathic) or secondary (to some known precipitating cause) FS.⁷ Factors that do not appear to affect the success rate are age, length or severity of symptoms, or diabetic status, but the latter does have a significant effect on the recurrence rate.^7,9

To date, we do not know the factors involved in causing the failure rate, or in recurrence in nondiabetic patients. Failure occurs even in patients who have had a well-documented release of the capsule (feeling of give at MUA followed by free ROM) and, therefore, it is speculated that it is caused by a rapid retightening of the shoulder capsule after the procedure.

Movement of the shoulder using supervised or nonsupervised physiotherapy activity immediately after MUA may therefore prevent failure, and may even prevent recurrence, but to date there appears to be no published work to support this theory. One study with a total of 41 patients found no difference in outcome of patients undergoing hydrodilation for FS with subsequent supervised physiotherapy versus a home exercise program.¹⁴ Physiotherapy alone may improve symptoms and ROM in patients with FS,^13,15–17 and indeed one study has suggested that it is effective with or without the MUA,¹⁸ but its effects are slow and prolonged treatment may be required.^19,20

A UK study comparing physiotherapy, MUA and arthroscopic release (UKFROST trial) looked at results at 12 months,²¹ and concluded that the difference between arthroscopic release and MUA was unlikely to be of clinical importance. It also determined that arthroscopic release was associated with higher risks, physiotherapy alone was more likely to need further treatment and MUA was, in fact, the most cost-effective treatment for FS even with 12 weeks of postprocedural physiotherapy.

A recent paper has shown that hydrodilation has a low rate of repeat intervention.¹⁰ In this study, physiotherapy with capsular stretching exercises was initiated within the first week with a subsequent routine 3-month course, although details of this are not provided. No explanation of why this method should result in a reduction in recurrence rate has been proposed. The results of hydrodilation may be similar to those achieved with MUA (although there have been no adequately powered randomised controlled trials to compare).

The study is adequately powered to show any difference in effect. The groups are matched for those factors that may affect the outcome or the recurrence rate (gender, primary or secondary FS or diabetic status).^7–9,22,23 A standardised method of MUA was used. There is a very low loss to follow-up rate, and the data are robust. There may be some factors that may skew this study, such as comorbidity, threshold for surgery, motivation for recovery and other unknowns. There is a significant difference in the time between diagnosis and treatment between the two groups, but the time to surgery has been shown to have no effect on the success of MUA and therefore is not a confounding factor.⁸

There was no difference in pre-MUA ROM between the groups. The mean OSS before the procedure was significantly greater for the MUA plus physiotherapy group than for the MUA and advice only group (23.0 vs 26.4) but this was recognised by the statistical methods used. Multiple surgeons performed MUA in Group 1 but a single surgeon in Group 2. However, all surgeons worked in the same team and the method of performing the MUA was standardised. We are unsure as to the degree of compliance of Group 1 patients with the physiotherapy advice but, even if there was no compliance at all, our results show that MUA without supervised physiotherapy is an effective treatment.

It is surprising that supervised physiotherapy postoperatively did not appear to reduce the recurrence rate. It is likely there may be confounding factors here, such as patient expectation, the threshold of the surgeon to offer repeat MUA or the ease or availability of the repeat MUA in a timely fashion. However, it appears to add weight to the argument that supervised physiotherapy may not prevent recurrence, and that it may be due to some underlying factors in the patient themselves (genetic or acquired), which are as yet unidentified.

The Royal College of Surgeons of England is advocating the development of elective surgery hubs to tackle the huge waiting lists caused by the COVID pandemic. This study may help commissioners to use physiotherapy resources efficiently following MUA for FS.

Conclusions

Following MUA for FS, a statistically significant increase in OSS was detected in patients receiving postprocedure physiotherapy compared with advice alone. There was no difference in recurrence rates.

However, the increase in OSS (3.2) due to the effects of physiotherapy is below the MCID, and both groups had a significant improvement in OSS (16.7 and 18.2, respectively) following MUA, raising questions regarding the relative importance of postprocedure physiotherapy in a resource-limited environment.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Conflicts of interest/competing interests

The authors declared no potential conflict of interest with respect to research, authorship and/or publication of this article.

Ethics approval

Ethical approval not required because this was a retrospective review of standard operative practice (SOP).