Ryans 2005.

Methods	Design: Parallel group, four‐arm, single blind randomised controlled trial (United Kingdom) Interventions: Physiotherapy (interferential current, proprioceptive neuromuscular facilitation, Maitland mobilizations and active exercise) plus glucocorticoid injection or glucocorticoid injection alone or physiotherapy plus placebo injection or placebo injection alone Sample size calculation: 20 participants per group were estimated to be needed based upon detecting a difference of 1.04 points on a 5‐point pain scale (SD=1.6) at 4 weeks at the 5% level of statistical significance with 82% power Analysis: Per protocol analysis Source of funding: Arthritis Research Campaign (non‐industry)
Participants	Number of participants: 80 (20 per group) Baseline characteristics: Physiotherapy plus glucocorticoid injection group: Mean (SD) age = 56.3 (6.4) years; Male:Female = 11:9 Mean (SD) duration of symptoms: 14.2 (4.4) weeks Glucocorticoid injection alone group: Mean (SD) age = 52.3 (9.3) years; Male:Female = 6:13 Mean (SD) duration of symptoms: 12.2 (5.3) weeks Physiotherapy plus placebo injection group: Mean (SD) age = 52.6 (7.7) years; Male:Female = 6:14 Mean (SD) duration of symptoms: 14.4 (4.4) weeks Placebo injection alone group: Mean (SD) age = 55.2 (9.4) years; Male:Female = 9:10 Mean (SD) duration of symptoms: 14.9 (3.7) weeks Inclusion criteria: 1. Aged 18 years or older 2. A painful shoulder, in the fifth cervical (C5) dermatome distribution, of more than four weeks and less than six months duration 3. Limitation of active and passive range of movement greater than 25% in abduction and external rotation compared with the other shoulder Exclusion criteria: 1. Pain was less than four weeks duration 2. Symptoms of more than six months duration 3. Had a previous intra‐articular injection or prior physiotherapy for this episode of shoulder pain 4. Presence of restriction of active and passive range of movement in external rotation only or glenohumeral abduction only 5. Had evidence of glenohumeral osteoarthritis on plain X‐ray 6. Had clinical evidence of a complete rotator cuff tear (i.e. positive drop‐off sign or weakness of the rotator cuff muscles) 7. Had clinical evidence of significant cervical spine disease, history of significant trauma to the shoulder or a history of inflammatory joint disease or of a cerebrovascular accident affecting the study shoulder 8. Had bilateral adhesive capsulitis 9. Had a contraindication to triamcinolone injection
Interventions	All participants were provided with 50x500mg paracetamol tablets with suggestions to take one or two tablets 4‐ to 6‐hourly as required for pain, taking no more than a maximum of eight tablets daily. All participants were also instructed by a physiotherapist in an identical home exercise programme using a video and home exercise instruction sheet Physiotherapy plus glucocorticoid injection (N=20) Components of physiotherapy intervention: ‐ Electrotherapy: Standardised interferential current ‐ Manual therapy: Maitland mobilizations which were progressed as the condition improved, and proprioceptive neuromuscular facilitation ‐ Supervised exercise: Active exercise therapy with gym equipment Dosage: Not reported Frequency of administration: Twice a week for four weeks (eight sessions) Provider: Physiotherapist Components of glucocorticoid injection: Injections of triamcinolone 20mg (1 ml) and normal saline 2 ml plus physiotherapy for four weeks. Injections were given (without imaging guidance) by a combined approach to the shoulder: half the solution (1.5 ml) was injected by an anterior approach and half (1.5 ml) by a lateral approach Glucocorticoid injection alone (N=20) The same injection method as described above was delivered Physiotherapy plus placebo injection (N=20) The same injection and physiotherapy method as described above was delivered, except that normal saline 3 ml was injected into the shoulder Placebo injection alone (N=20) The same injection method as described above was delivered, except that normal saline 3 ml was injected into the shoulder
Outcomes	Outcomes assessed at 6 and 16 weeks post‐randomisation Primary outcome: 1. Croft Shoulder Disability Questionnaire (0‐22 score range, where a score of 0 indicates no disability and a score of 5 and over represents significant disability) Secondary outcomes: 1. General health status using the SF‐36 (assessed at 16 weeks post‐randomisation only) 2. Passive and active range of motion in forward flexion, abduction, external rotation, internal rotation using a goniometer 3. Daytime pain at rest using a 100mm visual analogue scale 4. Global function using a 100mm visual analogue scale
Notes	*Outcome data fully reported only for these outcomes. No outcome data reported for other outcomes. Unpublished data regarding study design (required for risk of bias assessment) provided by trialist on request. Trial registered in ISRCTN but outcomes not provided at time of registration (http://www.controlled‐trials.com/ISRCTN25152388).
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Subjects were randomly allocated in permuted blocks of four using random number tables to one of four treatments. The randomization process took place in the hospital pharmacy department. Allocations were placed in sealed envelopes which were opened by the physiotherapist teaching the home exercise programme" Comment: An adequate method was used to generate the allocation sequence
Allocation concealment (selection bias)	Low risk	Comment: See quote above. An adequate method was used to conceal the allocation sequence
Blinding of participants and personnel (performance bias) All outcomes	High risk	Quote: "Injections were provided in opaque syringes, and the investigator measuring outcomes (IR) was not present at the time of randomization or injection and was blinded to all study interventions. Both patients and the physiotherapist were blinded to the nature of the injection. Clearly, it was impossible to blind subjects regarding physiotherapy but subjects were asked not to reveal if they were having physiotherapy treatment." Comment: Participants and personnel were blind to the injection component of the intervention, but not the physiotherapy component. Participants may have had different expectations about the benefits of each intervention
Blinding of outcome assessment (detection bias) Self‐reported outcomes	High risk	Comment: Participants self‐reported pain, general health status and function, and were not blind to whether they had received physiotherapy or not. Participants may have had different expectations about the benefits of each intervention
Blinding of outcome assessment (detection bias) Objective outcomes	Low risk	Quote: "Injections were provided in opaque syringes, and the investigator measuring outcomes (IR) was not present at the time of randomization or injection and was blinded to all study interventions. Both patients and the physiotherapist were blinded to the nature of the injection. Clearly, it was impossible to blind subjects regarding physiotherapy but subjects were asked not to reveal if they were having physiotherapy treatment." Comment: Assessors of objective outcomes were blind to treatment.
Incomplete outcome data (attrition bias) All outcomes	High risk	Quote: "Eighty subjects were recruited and randomly assigned to four groups. One subject was randomized twice and another failed to attend for intervention after randomization; 78 subjects were therefore available for analysis. Twenty subjects were enrolled in Group A (steroid injection and physiotherapy), 19 in Group B (steroid injection and no physiotherapy), 20 in Group C (placebo injection and physiotherapy) and 19 in Group D (placebo injection and no physiotherapy). Six subjects did not return for all follow‐up visits: three in Group A, one in Group B, one in Group C and one in Group D. Fifteen subjects withdrew from the study due to failure of the study treatment. Six patients withdrew from Group B, three from Group C and six from Group D" Quote: "We also looked to see if there were significant differences in numbers dropping out in each group due to failure of treatment. Significantly more patients dropped out in Group D (placebo injection and no physiotherapy) and in Group B (steroid injection and no physiotherapy (Pearson chi‐square = 8.72, P=0.033). No subjects dropped out of Group A (steroid injection and physiotherapy)." Comment: The was differential drop‐out across the groups and the reasons appear to be related to the treatments received
Selective reporting (reporting bias)	High risk	Quote: "Secondary outcome measures were...range of movement as measured by passive external rotation. External rotation was chosen as the indicator range of movement as restriction in this range has been described as the most severely restricted plane of movement in shoulder capsulitis" Quote: "Analysis of improvement in the range of movement in abduction and internal rotation (thumb–C7 distance) revealed no significant association with either steroid injection or physiotherapy." Comment: Trialists reported measuring passive and active range of motion (forward flexion, abduction, external rotation, internal rotation) using a goniometer. However, outcome data was only reported for passive external rotation. The decision not to report outcome data for the other measures of range of motion appears to be related to the statistical significance of the results
Other bias	Low risk	Comment: No other sources of bias identified