At-Home Shoulder Exercises or Supervised Physical Therapy for Adhesive Capsulitis: A Retrospective Comparative Cohort Study

Abstract

Background:

The purpose of this study was to determine if the clinical outcomes of patients with adhesive capsulitis were influenced by the type of shoulder exercise regimen prescribed.

Methods:

This was a retrospective, comparative cohort, therapeutic outcomes study of patients with primary or secondary systemic adhesive capsulitis who were initially treated with either a doctor-directed home-based exercise program (HEP) or the HEP supplemented with supervised physical therapy (SPT). Study participants were treated by one of two orthopaedists at a single institution between 2012 and 2022. Demographic and disease-specific data extracted from electronic medical records during the period of observation were compiled in an electronic database. Primary outcomes included shoulder range of motion (ROM) measures and treatment failures leading to operative intervention. Data were analyzed using the Friedman test and the Mann-Whitney U exact test.

Results:

There were 124 patients who met eligibility criteria, 81 in the HEP cohort and 43 in the SPT group. Intra-cohort comparisons showed significant increases in shoulder ROM in all planes between baseline and follow-up encounters (p<0.001), indicating clinical improvement with either shoulder rehabilitation method. However, inter-cohort comparisons demonstrated shoulder flexion and abduction were significantly greater in the HEP group than in the SPT group at both 8-week and final follow-up evaluations (p<0.001). While there were no treatment failures in the HEP cohort, 10 of 43 cases (23%) failed the SPT protocol, wherein an unconscious framing bias regarding treatment expectations and early operative intervention may have contributed to the high failure rate. Good clinical outcomes were obtained in all 10 cases with shoulder manipulation under anesthesia, arthroscopic global capsulotomy, and continued supervised physiotherapy.

Conclusion:

Patients with adhesive capsulitis should be treated initially with at-home shoulder stretching exercises, reserving more costly supervised physical therapy or operative intervention for recalcitrant cases.

Level of Evidence: III

Introduction

“Adhesive capsulitis” is a shoulder condition of uncertain etiology characterized by the insidious onset of pain and progressive loss of active and passive motion without an underlying intrinsic shoulder disorder.¹ The term was coined in 1945 by Neviaser² who found that the pathology involves a thickening and contracture of the glenohumeral joint capsule which he believed was adherent to the humeral head. Despite the absence of intra-articular adhesions found in the arthroscopic era, the misnomer “adhesive capsulitis” endures in present terminology. Also known as “frozen shoulder,”³ primary adhesive capsulitis must be distinguished from glenohumeral joint stiffness secondary to shoulder trauma, surgery, or immobilization.⁴

The disease is initiated by an inflammatory process leading to recruitment of fibroblasts which are transformed into myofibroblasts, ultimately creating scar bands within the joint capsule.^2,5 Although the exact pathophysiology remains unclear, the essential lesion of adhesive capsulitis is contracture of the rotator cuff interval and the coracohumeral ligament^6,7 which then act as a checkrein restricting shoulder external rotation.⁸ As the disease progresses, contracture of the axillary recess ensues, thereby limiting shoulder motion globally. Histologic examination of the involved tissue typically demonstrates capsular fibrosis, hyalinization, and fibrinoid degeneration.⁹

While the incidence of adhesive capsulitis is an estimated 2-5% in the general population,⁴ the disease occurs much more frequently in women and among those with certain systemic conditions such as diabetes mellitus.^10,11 In a recent epidemiologic study of a large database,¹² the median age of nearly 166,000 patients with adhesive capsulitis was 58 years, 67% were women, 44% had diabetes, and 30% had thyroid disorder. In addition, a recent 500,000-patient database genomic study found the genetic risk of adhesive capsulitis was similar to that of diabetes and hypothyroidism, implicating the Wnt pathway and cellular proliferation response in the pathomechanism of adhesive capsulitis.¹³

The natural history of adhesive capsulitis, as originally described by Reeves,¹⁴ includes a continuum of three clinical stages. The painful inflammatory or “freezing” phase (months 2 to 9) is characterized by increasing shoulder pain and loss of motion. In the progressive stiffness or “frozen” phase (months 3 to 12), shoulder motion becomes limited in all planes, but pain subsides as shoulder inflammation abates. The resolution or “thawing” phase is marked by gradual improvement in both shoulder pain and ROM (months 12 to 24 or more). Although idiopathic frozen shoulder is often described as a self-limited disease which may resolve without specific treatment,^15,16 a recent systematic review calls the dogma of spontaneous recovery into question.¹⁷ To the contrary, many patients with adhesive capsulitis experience a persistent deficit of shoulder motion for years after disease onset, albeit the loss of motion is usually not associated with functional impairment.^18,19

Gentle passive stretching exercises for frozen shoulder have been the cornerstone of nonoperative treat-ment,^20–23 but clinical improvement is not universal. One study of nearly 2200 patients demonstrated that 16.5% failed conservative treatment and required surgery.²⁴ The present study sought to find differences in the clinical outcomes of adhesive capsulitis patients treated with an at-home exercise program compared with a cohort managed with supervised physical therapy. We hypothesized that there would be no significant differences in key outcome measures as a function of the shoulder exercise regimen.

Methods:

This was a retrospective comparative cohort study of patients with primary or secondary systemic adhesive capsulitis treated with either a home-based stretchingexercise program (HEP) or a supervised physical therapy (SPT) regimen. One of the senior authors (HAM) employed a HEP described as “orthotherapy,”²⁰ whereas another (BFH) recommended a SPT regimen in addition to an at-home shoulder stretching routine. Formal physical therapy in the latter group was prescribed to assure patients were performing stretching exercises properly and to assuage patient concerns that all measures were being employed to expedite their recovery. This research study was approved by our university institutional review board.

Diagnosis

Diagnostic criteria for adhesive capsulitis include (1) history of painful restricted shoulder motion without antecedent trauma, (2) physical exam demonstrating global shoulder stiffness without loss of intrinsic strength, and (3) plain radiographs of the shoulder showing normal glenohumeral joint interval without periarticular abnormalities except for disuse osteopenia.²⁵ Shoulder examination in our study was usually conducted with the patient in the seated position. Although the physical exam typically revealed limited shoulder motion in all planes (forward flexion, abduction, cross-body adduction, external and internal rotation with the arm at the side), documentation in the medical record usually included only three measures: forward flexion in the sagittal plane, abduction in the scapular plane, and external rotation with the arm at the side. All patients had standard orthogonal radiographs of the affected shoulder including Grashey or true anteroposterior (AP), axillary lateral, supraspinatus outlet, and 30° caudal tilt AP projections.

Patient Screening and Study Eligibility Criteria

Patients for this study were identified by searching our office electronic database using the International Classification of Diseases (ICD) codes for adhesive capsulitis (ICD-9 code 726.0, ICD-10 code M75.0x). Eligible patients were adults >18 years of age with the pertinent ICD codes who had initial nonoperative treatment by the senior authors between 2012 and 2022 during the six-year periods before they retired from clinical practice.

Excluded from study were patients with (1) intrinsic problems such as antecedent shoulder trauma or surgery or concurrent shoulder pathology such as rotator cuff tear or (2) extrinsic conditions such polyarthritis or neurologic disease producing shoulder pain. Those with insufficient documentation of pre- or post-treatment shoulder range of motion (ROM) measures or with unrestricted global ROM were also excluded. However, patients with diabetes mellitus and/or thyroid disease were not excluded, so that this study incorporated those with idiopathic and secondary systemic adhesive capsulitis.¹

Treatment Protocol

Conservative treatment comprised three phases. First, shoulder rest, anti-inflammatory medication, and moist heat were recommended. Second was the passive shoulder motion phase of rehabilitation (Table 1), prescribed for all patients in both HEP and SPT cohorts. In addition to engaging in passive shoulder stretching exercises at home, physiotherapy for those in the SPT group was supervised by a physical therapist or physical therapy assistant once or twice weekly. The goal of this phase was to achieve full painless shoulder ROM. Third, when patients regained about 90% of motion, shoulder strengthening was added to the regimen. Patients were informed beforehand that recovery from frozen shoulder could take from weeks or months to one year or more and were encouraged to be patient with the rehabilitation process.

Table 1.

Passive Shoulder Stretching Exercises for Adhesive Capsulitis

Codman Pendulum Stand adjacent to a chair or table and bend slightly at the waist. Support your body weight by leaning on the chair or table with the opposite arm. Let the full weight of the affected arm hang freely. Gently swing the affected arm forward and back, side to side.

Wall Climb Face a wall with hands out in front of body, touching the wall. Gradually use fingers to climb the wall, slowly raising the arms. At the peak, hold for 15-30 seconds.

Posterior Shoulder Capsular Stretch Bring affected arm across body to feel the stretch in the back of the shoulder. Hold for 15-30 seconds.

Passive Forward Flexion With your opposite arm or the help of your therapist, gently raise the affected arm up in front. Start with your affected arm at your side and passively bring your hand to mouth, then forehead, then top of head.

Supine Overhead Stretching Begin in a supine position (lying flat on bed). Grip a 3-foot stick with both hands. Raise the stick slowly overhead as tolerated. Hold for 5 seconds.

Stick Exercises Using a 3-foot stick, cane or yard stick, perform passive range of motion in flexion, extension, abduction, external and internal rotation.

Overhead Pulley Secure overhead pulley in doorway. Grasp both handles. At first, pull down on the normal or unaffected side while the stiff shoulder is relaxed (passive motion). Progress to active motion of the stiff shoulder.

Notes: All shoulder motion must be painfree. All patients in the study were instructed to perform this exercise regimen for 10-15 minutes, at least 3 times daily. In addition, patients in the supervised physical therapy (SPT) group were guided by a physical therapist 1-2 times weekly.

If there was no symptomatic relief and/or if shoulder ROM did not objectively improve with this conservative approach, then patients were offered the option of continued physiotherapy or shoulder manipulation under anesthesia and arthroscopic global capsulotomy. One of the senior authors (HAM) required a minimum three months of passive shoulder exercises before surgical intervention, while the other (BFH) indicated patients for surgery whenever shoulder motion plateaued following three successive SPT sessions, regardless of treatment duration. This escalation of care was considered failure of conservative treatment.

Data Collection and Outcome Measures

Patient demographic data including age, body mass index (BMI), biological sex, medical co-morbidities, smoking status, and affected side were extracted from electronic medical records. Longitudinal shoulder ROM data included forward flexion in the sagittal plane, abduction in the scapular plane, and external rotation with the arm at the side. These measures on the affected side were recorded (1) before commencing treatment to establish a baseline, (2) at eight weeks (or sooner) after initiating therapy, and (3) at the final follow-up encounter. Motion measurements were made by surgeons or midlevel assistants using hand-held goniometers.

Primary outcome measures included shoulder ROM, duration of treatment, and treatment failures leading to operative management. Patients indicated for operative treatment and categorized as physiotherapy failures also had postoperative shoulder ROM measures included in the database. Corticosteroid injections in both treatment groups and therapy sessions in the SPT cohort were enumerated. However, no long-term clinical evaluations of patients in this study were conducted.

Statistical Analysis

All pertinent data were entered into the Research Electronic Data Capture (REDCap) system^26,27 for permanent storage and analysis. Categorical variables were summarized as frequencies and percentages. Continuous variables were assessed for normality and reported as means and standard deviations or medians and interquartile ranges. No estimates were made for missing data noted in some follow-up encounters. Where data were sparse, between-group differences were assessed with the Mann-Whitney U exact test. Statistical tests were conducted in IBM SPSS Statistics, version 29 using two-sided tests and alpha level of significance of 0.05.

Results:

A total of 124 patients met eligibility criteria, 81 in the HEP cohort and 43 in the SPT group. Most patients in this study had shoulder pain and progressive shoulder stiffness consistent with the second or “frozen” phase of adhesive capsulitis. Comparison of demographic parameters (Table 2) showed those in the HEP cohort were significantly older than their SPT counterparts by about 5 years (p=0.009) and had a higher prevalence of hypothyroidism, but the groups were otherwise similar.

Table 2.

Patient Demographics by Therapy Type

Characteristic	SPT	HEP	X2 (df)	p
	n=43 (%)	n=81 (%)
Age, yrs, mean (sd)	54.9 (7.6)	59.4 (9.6)	2.6 (122)	0.009**
BMI, calculated, mean (sd)	27.1 (6.3)*	27.9 (6.0)	0.54 (95)	0.593**
Biological Sex
Male	16 (37.2)	28 (34.6)	0.086 (1)	0.770
Female	27 (62.8)	53 (65.4)
Co-Morbidities
Diabetes mellitus	11 (25.6)	21 (26.3)	0.002 (1)	0.967
Hypothyroidism	0 (0)	8 (9.9)	4.54 (1)	0.050
Symptomatic shoulder
Right	20 (46.5)	34 (42.0)	0.235	0.628
Left	23 (53.5)	47 (58.0)

Notes: SPT, supervised physical therapy; HEP, home exercise program; BMI, body mass index. X² (df): Chi-square test of independence (degrees of freedom). Fisher’s exact test was reported when data were sparse.

^*BMI was missing for 27 patients in the SPT group.

^**Independent t-test for equality of means.

Comparison with bolded p-value met statistical significance, p<0.05.

Clinical outcome data are shown in Table 3. The Friedman test was used to analyze the shoulder range of motion data within each cohort. There were statistically significant increases in median shoulder flexion, abduction, and external rotation between baseline and both follow-up intervals within the HEP and SPT groups (p<0.001), indicating marked clinical improvement with either shoulder rehabilitation protocol.

Table 3.

Clinical Data Including Shoulder Range of Motion at Baseline, After 8 Weeks of Treatment, and at Final Follow-Up

Characteristic	SPT		HEP
	n	median (IQR)	n	median (IQR)	p-value*
Duration of pretreatment symptoms, days	39	150 (63, 210)	17	150 (120, 270)	0.531
Duration of treatment, days	43	70 (42, 124)	81	49 (42, 117)	0.358
Flexion ROM, degrees
Baseline	43	110 (100, 128)	81	120 (90, 150)	0.124
8 weeks	39	126 (110, 148)	74	160 (120, 170)	<0.001
Final	43	135 (120, 152)	81	165 (145, 170)	<0.001
Friedman test		p <0.001		p <0.001
Abduction ROM, degrees
Baseline	43	97 (75, 138)	81	95 (80, 150)	0.574
8 weeks	39	140 (100, 154)	75	160 (100, 170)	<0.001
Final	43	150 (130, 160)	81	165 (135, 170)	<0.001
Friedman test		p <0.001		p <0.001
External Rotation ROM, degrees
Baseline	43	24 (17, 35)	81	20 (10, 45)	0.992
8 weeks	39	35 (19, 45)	75	45 (20, 60)	0.138
Final	43	40 (30, 55)	81	45 (25, 60)	0.252
Friedman test		p <0.001		p <0.001

Notes: SPT, supervised physical therapy; HEP, home exercise program; IQR, interquartile range; ROM, range of motion.

^*Mann-Whitney U exact test.

Bolded p-values were statistically significant, p<0.05.

The Mann-Whitney U exact test was used to compare the compiled data in the SPT group to the corresponding values in the HEP cohort. There was no difference in the duration of pretreatment symptoms between the groups, although many patients in the HEP group did not have pre-treatment length of symptoms documented in their medical record. The duration of treatment in the SPT group was, on average, three weeks longer than in the HEP cohort but this difference was not significant. In addition, those in the SPT group had a median of 12 (IQR 8, 20) supervised therapy sessions during the course of treatment, indicating substantially greater healthcare cost compared with the nominal expense of one-time patient education in the HEP group.

While there were no significant differences in shoulder flexion, abduction, or external between the two cohorts at baseline, measured flexion and abduction were significantly greater in the HEP group than in the SPT group at both the eight-week and final follow-up. For example, at the 8-week follow-up, median shoulder flexion in the HEP group was 160°, but only 126° in the SPT group (p<0.001). At the final follow-up visit, median abduction was 15° greater in the HEP cohort compared with the SPT group (p<0.001). No significant differences in external rotation measures were found at either the 8-week or final clinical checks.

Importantly, ten treatment failures in 43 cases (23%) were found in the SPT group after a median 61 (IQR 35, 87) days. However, there were no failures of conservative treatment in the HEP cohort. All SPT failures underwent shoulder manipulation under anesthesia and arthroscopic global capsulotomy and had prompt resumption of supervised shoulder stretching and ROM exercises after surgery. No intra- or post-operative complications in this subgroup were noted.

Subgroup analysis (Table 4) revealed that the 10 who failed supervised physical therapy were, on average, younger than those in the successfully treated subgroup and all but one were women. There was not a higher proportion of diabetics among those requiring surgery. Although the baseline shoulder ROM of the SPT failures was less than the SPT successes, the final shoulder ROM of the SPT failures after surgical intervention and further rehabilitation was better than their counterparts, with the increase in shoulder flexion being statistically significant (p=0.024). All 10 were judged to have a good clinical result following surgical intervention and continued supervised physical therapy. However, on average, those in the failed SPT subgroup required twice the number of therapy sessions (p<0.001) and 2.4 times the duration of treatment (p<0.001) compared with the subgroup successfully treated with physiotherapy alone.

Table 4.

Comparison of Supervised Physical Therapy Outcome Successes with Outcome Failures

Characteristic	SPT Success (n=33)	SPT Failure (n=10)	p
Age, yrs, mean (sd)	55.7 (7.6)	52.4 (7.3)	0.233*
Biological Sex, n (%)
Male	15 (45.5)	1 (10.0)	0.063**
Female	18 (54.5)	9 (90.0)
Diabetes mellitus, n (%)	9 (27.3)	2 (20.0)	>.999**
Symptomatic shoulder, n (%)
Right	14 (42.4)	6 (60.0)	0.473**
Left	19 (57.6)	4 (40.0)
Baseline ROM, degrees, median (IQR)
Flexion	110 (100, 128)	108 (102, 128)	0.983
Abduction	102 (77, 145)	92 (75, 108)	0.510
External rotation	28 (18, 36)	18 (8, 26)	0.094
Final ROM, degrees, median (IQR)
Flexion	135 (120, 148)	151 (150, 155)	0.024
Abduction	146 (128, 157)	160 (142, 162)	0.058
External rotation	38 (30, 48)	45 (30, 55)	0.500
Total SPT sessions, median (IQR)	12 (8, 14)	24 (14, 35)	<0.001
Duration of treatment, days, median (IQR)	56 (42, 91)	137 (124, 177)	<0.001

Notes: SPT, supervised physical therapy; ROM range of motion; IQR, interquartile range.

p-values:

^*t-test

^**Fisher's exact test for sparse data; otherwise, Mann-Whitney U exact test.

Bolded p-values are statistically significant, p<0.05.

Our study was not well controlled for glenohumeral or subacromial corticosteroid injections which were administered sparingly to relieve acute shoulder pain and to augment the shoulder exercise regimens. In the 43-patient SPT group, two received an injection at the initial office visit, while five others had six injections at follow-up visits. In the larger 81-patient HEP cohort, two patients received an injection initially, while five other patients were administered injections at subsequent office visits.

Additional subgroup analysis of the outcome failures revealed that more supervised physical therapy sessions and a longer duration of treatment were required after surgical intervention than before surgery (Table 5). However, these differences were not statistically significant.

Table 5.

Supervised Therapy Sessions and Duration of Treatment Before and After Surgery Among Outcome Failures

Characteristic	Before Surgery	After Surgery	p-value*
Therapy sessions, median (IQR)	7 (3, 16)	14 (5, 24)	0.145
Duration of treatment, days, median (IQR)	61 (35, 87)	83 (53, 101)	0.285

Note: IQR, interquartile range.

^*Wilcoxon signed ranks exact test.

Discussion

This retrospective study demonstrated that improved shoulder motion may be achieved in adhesive capsulitis with either doctor-directed home-based exercises or with supervised physical therapy. However, contrary to our null hypothesis, patients in the HEP cohort, had significantly greater shoulder flexion and abduction at 8-week and final follow-up encounters compared with those in the SPT group, indicating more favorable short-term outcomes with home-based exercises. Moreover, most of the improvement in shoulder flexion and abduction in the HEP group occurred during the first eight weeks of treatment (Table 3). Forty-four of 81 patients (54%) in the HEP group improved substantially within eight weeks and required no further follow-up. These patients were advised to continue shoulder exercises independently as needed to maximize their shoulder ROM.

Also counter to our null hypothesis, there were 10 treatment failures in the SPT cohort but no failures in the HEP group. As others have speculated,^20,28 our experience suggests that physiotherapy failures may be caused by too aggressive and painful stretching of the shoulder capsule which exacerbates the inflammatory process during the late freezing phase or early frozen phase of the disease. This may be true despite admonitions for therapists to avoid maneuvers evoking shoulder pain. On the other hand, patients in the HEP cohort were apparently more likely to perform passive shoulder stretches in the pain-free range, leading to a successful outcome. Of course, patient compliance with the unsupervised home-based exercise program cannot be confirmed.

Ad hoc discussions among investigators after data collection revealed two key differences in how the senior authors framed patient treatment expectations. First, those in the HEP cohort met with a physician assistant during the initial office visit to review details of the home-based exercise program emphasizing that recovery could take months to years, whereas patient education in the SPT group was delegated to physical therapists. Second, the SPT cohort was initially informed that operative intervention would be considered in the event of shoulder ROM failing to improve whatsoever or somewhat improving but then plateauing during three successive physical therapy sessions. Thus, an unconscious framing bias²⁹ may have been established in the SPT patients predisposing them to choose surgery when presented with the opportunity to do so. On the other hand, when HEP patients struggled to regain shoulder motion for three months or more and were offered operative intervention versus continuing the at-home exercise program, they consistently chose the nonoperative option, being biased to expect a longer recovery. This finding highlights the importance of setting evidenced-based patient expectations at the index office encounter.

There are few publications comparing two shoulder exercise regimens for adhesive capsulitis, as in the present study. In a prospective study of 77 patients with frozen shoulder treated with supervised passive stretching and manual mobilization or with pendulum exercises and activities of daily living (ADLs) in the pain-free range, Diercks and Stevens²⁸ found that 89% in the latter group had normal or near normal shoulder function after 2 years compared with 63% in the former group. There were no treatment failures in either group. At one-year follow-up in a 75-patient randomized controlled trial (RCT), investigators in the United Kingdom³⁰ reported that shoulder ROM and other outcome measures were significantly improved by supervised physiotherapy compared with home exercises. No treatment failures were noted in either cohort studied. Robinson et al.³¹ conducted a 41-patient RCT of supervised physiotherapy versus a home exercise program after hydrodistention of the glenohumeral joint for primary frozen shoulder and found no significant difference in clinical outcomes between the two groups at one-year follow-up. Only one patient in the home exercise group failed treatment and required an arthroscopic procedure.

Most recently, in a 61-patient RCT at one-year followup, Martin et al.³² showed similar patient-reported outcome measures between patients in the physiotherapy group compared with those randomized to the “watchful waiting” cohort who were permitted to use the affected shoulder for ADLs only. The study also concluded that patients treated with supervised physical therapy incurred significantly higher direct costs and imposed a greater economic burden on the healthcare system than their counterparts managed with benign neglect. No treatment failures were reported in either cohort. A noteworthy comment was that 65% of those approached for enrollment declined to participate in the study, most preferring supervised physical therapy,³² despite understanding the absence of clinical evidence for its efficacy, as is well delineated elsewhere.¹⁶ In the present study, we observed strong patient preferences in the SPT cohort, both for formal physical therapy initially, and subsequently for operative intervention if therapy failed, in the hope of expediting recovery. Of course, the potential risks, perceived benefits, and associated costs of physiotherapy and surgery must be discussed with patients and carefully considered in all treatment decisions.

There were several study limitations including the retrospective design, relatively small sample size, and lack of long-term clinical follow-up. In addition, corticosteroid injections were not controlled and, therefore, were a confounding factor. Assessment of glenohumeral joint ROM using hand-held goniometers introduced the possibility of device-related measurement errors, and participation of multiple examiners over many years created the possibility of inter-observer measurement errors. Some patients, particularly those unwilling to be patient with rehabilitation, may have sought alternative treatment elsewhere. Over 50% in the HEP cohort and 40% in the SPT group had only one follow-up office visit and may have been treated by other healthcare providers for recurrent symptoms. Finally, framing bias regarding patient treatment expectations may have influenced the study outcome.

On the basis of our findings, we recommend that patients with adhesive capsulitis be started on a home exercise program or “orthotherapy” as described by Rockwood,²⁰ reserving more costly supervised therapy or operative intervention for recalcitrant cases. Oral antiinflammatory medications and corticosteroid injections may be used to augment the exercise program. Although this study was not designed to estimate the duration of nonoperative treatment, the literature suggests that a minimum of three²¹ to six^4,16 months of conservative measures should precede surgical intervention.

All patients, but particularly those with work and/ or athletic restrictions, must be patient with the rehabilitation process and be persuaded that “less is more” regarding the treatment of adhesive capsulitis. A minority of patients in both cohorts of our study had residual shoulder stiffness at last follow-up visit, confirming that some will take longer to improve than others. If nonoperative treatment fails, a good clinical outcome may be obtained with shoulder manipulation under anesthesia, arthroscopic global capsulotomy, and continued physiotherapy, albeit with significantly greater treatment duration and healthcare cost.

Conclusion:

Although shoulder ROM in primary or secondary systemic adhesive capsulitis significantly improved with either an at-home shoulder stretching exercise program (HEP) or with supervised physical therapy (SPT), those in the HEP group had significantly greater shoulder flexion and abduction at 8-week and final follow-up visits than the SPT group. Moreover, patients in the HEP cohort had an average of three weeks less treatment and incurred no treatment failures. On the other hand, 10 of 43 patients (23%) failed SPT, in part owing to an unconscious framing bias regarding treatment expectations and early operative intervention. Based on our findings, patients with adhesive capsulitis should be treated initially with at-home shoulder stretching exercises, reserving more costly supervised therapy or operative intervention for recalcitrant cases.